# freddy — full content index
# Last updated: 2026-10-30

## Every screenless fitness tracker worth buying in 2026, scored and tiered
URL: https://freddy.coach/articles/best-screenless-fitness-trackers
Seven screen-free bands and rings scored on eight things that matter, from sleep and recovery to battery, price and who can read your data.
Screenless wearables are the fastest-growing corner of fitness tracking. No display, no notifications, no wrist that lights up at dinner. Just a sensor that sits on your wrist, arm or finger and reports what your body did. WHOOP made the format famous. In 2025 and 2026 Polar, Garmin, Amazfit and Google all shipped their own, and the smart rings from Oura and Ultrahuman compete for the same buyer. This guide scores the seven worth buying today on eight dimensions, puts them into tiers, and shows the facts behind each score. It is written for someone who has narrowed the choice to "no screen" and wants to know which one, and why. How we scored Each device gets a score from 1 to 10 on eight dimensions: sleep tracking, recovery and HRV, activity tracking, battery and comfort, sensor technology, extra features, data openness, and value over three years. The scores come from published validation studies (with the hardware generation each study tested), vendor specifications, and the consensus of established reviewers. We did not test the devices by hand. No vendor paid for, saw or reviewed this ranking. The tier is an editorial verdict that the scores inform, not a weighted sum. The scores are published so you can see the reasoning and disagree with it. The scores Scores out of 10 Dimension WHOOP 5.0 Garmin CIRQA Oura Ring 5 Polar Loop Google Fitbit Air Ultrahuman Ring PRO Amazfit Helio Strap Sleep tracking 8 7 9 6 7 6 6 Recovery and HRV 9 8 8 7 7 6 6 Activity tracking 7 9 5 6 6 4 7 Battery and comfort 9 8 7 7 7 9 8 Sensor technology 9 7 8 4 7 7 6 Extra features 8 6 8 3 7 6 5 Data openness 7 6 5 9 5 8 5 Value (3 years) 4 8 4 8 9 6 9 The tiers God tier WHOOP 5.0. The most complete recovery system in the category, two weeks per charge, and a band you charge while you wear it. The catch is the price: you pay every year, forever. Garmin CIRQA. Garmin's training science in a $199.99 band with no subscription. The best activity tracking here, and it broadcasts heart rate to your bike computer. Great Oura Ring 5. The best-studied sleep tracker you can buy, now in a much smaller ring. The membership is needed for full insights and for any third-party access to your data. Polar Loop. The most open data story in this list, no subscription, and Polar's long-trusted Nightly Recharge. The sensor set is the thinnest here. Google Fitbit Air. $99.99, 12 grams, AFib alerts and a free core feature set. The cheapest way into this category that does not feel cheap. Good, with caveats Ultrahuman Ring PRO. Fifteen days per charge and no subscription, but supply is limited and the US version has no AFib feature. Amazfit Helio Strap. Strong value and strength-training features at $99.99, but reviewers find its automatic workout detection unreliable. Skip for now None this round. Every device in the list does its core job. The caveats above are the reasons a device sits lower, not reasons to avoid it. Sleep tracking Sleep is the main reason most people buy a screenless tracker, and it is the dimension with the most published evidence. That evidence is thinner than the marketing suggests: no independent study has yet tested the current generation of any device here. Device Sleep stages Automatic sleep detection WHOOP 5.0 Yes Yes, with naps Garmin CIRQA Yes Yes, with naps Oura Ring 5 Yes Yes, naps of 15 minutes to 3 hours Polar Loop Yes Yes; naps not documented Google Fitbit Air Yes Yes; naps not documented Ultrahuman Ring PRO Yes Yes, with naps Amazfit Helio Strap Yes Yes, naps over 20 minutes Oura has the deepest research record. A 2024 study in Sleep Medicine found 79% agreement with polysomnography on four-stage classification across 96 participants ( source ). A 2025 study of a clinical sleep-lab population found 53% four-stage accuracy ( source ). Both tested the Gen3 ring, not the Ring 5. WHOOP 's main validation, Miller et al. 2020, found 89% agreement on sleep versus wake ( source ). A 2025 six-device study found it classified about 62% of REM correctly ( source ). Both tested WHOOP 4.0. Garmin , Google and Polar have long sleep-tracking histories on their watches, but no independent study of these bands yet. Ultrahuman and Amazfit have no independent sleep-stage validation that we could find. Recovery and HRV Every device here measures heart rate variability with the same method, RMSSD. They differ in when they measure it and in what they build on top of it. Device Documented HRV metric WHOOP 5.0 RMSSD Garmin CIRQA RMSSD Oura Ring 5 RMSSD Polar Loop RMSSD Google Fitbit Air RMSSD Ultrahuman Ring PRO RMSSD Amazfit Helio Strap RMSSD WHOOP measures HRV in the last slow-wave sleep period of the night and builds its daily Recovery score on it. Recovery is the product, and it shows. Garmin averages overnight RMSSD into HRV Status against your own three-week baseline, and feeds it into Training Readiness and Body Battery. Oura turns it into a Readiness score. Polar 's Nightly Recharge combines HRV with sleep into one morning reading. Google reports nightly HRV from your longest sleep. Amazfit rolls it into its BioCharge score. Ultrahuman changed its sleep algorithm in September 2026. Its own study of 1,026 members found the update lowered sleep HRV by 33.8% at the median ( source ). If you compare months of Ultrahuman data, expect a step in the line. The same is true of any device: HRV numbers from different brands are not comparable with each other, even when they use the same method. Activity tracking None of these devices has GPS. For a run with pace and distance, each one borrows your phone's GPS. The differences are in what they detect on their own. Device Automatic activity detection Strength rep counting Built-in GPS Heart-rate broadcast WHOOP 5.0 Yes, 45 types No No Yes, Bluetooth Garmin CIRQA Yes Yes No Yes, ANT+ and Bluetooth Oura Ring 5 Yes No No No Polar Loop Yes No No No Google Fitbit Air Yes Not documented No Not documented Ultrahuman Ring PRO Partial: it detects, you confirm No No No Amazfit Helio Strap Partial: reviewers call it unreliable Yes, 25 movements No Yes, Bluetooth Garmin wins this dimension clearly. The CIRQA detects workouts on its own, counts strength reps, has more than 80 sport types and Garmin's full training-load metrics, and broadcasts heart rate over ANT+ and Bluetooth to a bike computer or a gym machine. Amazfit counts reps across 25 movements and has a HYROX race mode, but its automatic detection is the weak point. WHOOP detects 45 activity types and measures strain well, but it does not count reps. The rings are weaker here by design: a ring on your finger is a poor place to measure a barbell. Battery and comfort Device Battery (vendor claim) WHOOP 5.0 14 days Garmin CIRQA 10 days Oura Ring 5 6 to 9 days Polar Loop 8 days Google Fitbit Air 7 days Ultrahuman Ring PRO 15 days Amazfit Helio Strap 10 days WHOOP and Ultrahuman lead. WHOOP also lets you slide a battery pack onto the band and charge without taking it off, so in practice it never leaves your wrist. Garmin and Amazfit last about ten days. The Fitbit Air weighs 12 grams with the band, and a five-minute charge gives a day. The Oura Ring 5 is 40% smaller than the Ring 4, and that smaller body costs it some battery: Oura now says 6 to 9 days, depending on ring size. Sensor technology Device SpO2 Skin temperature ECG WHOOP 5.0 Yes Yes WHOOP MG only Garmin CIRQA Yes Yes No Oura Ring 5 Yes Yes No Polar Loop No Partial: developer SDK only No Google Fitbit Air Yes Yes No Ultrahuman Ring PRO Yes Yes No Amazfit Helio Strap Yes Sensor listed; feature not documented No WHOOP is the only device here with an ECG, and only on the WHOOP MG hardware in the top membership tier. Every other device covers SpO2 and skin temperature, except Polar , which has neither as a user feature. Oura claims its new sensor design gives about 12% more accurate overnight HRV than the Ring 4. That is a vendor claim, not yet independently tested. Extra features Device AFib notifications Blood-pressure feature Women's health WHOOP 5.0 WHOOP MG only WHOOP MG: Blood Pressure Insights Yes Garmin CIRQA No No Yes Oura Ring 5 No Partial: Blood Pressure Signals, a wellness feature Yes, with membership Polar Loop No No No built-in feature Google Fitbit Air Yes No Yes Ultrahuman Ring PRO No (US) No Yes, free Amazfit Helio Strap No No Not documented The Fitbit Air is the only band under $200 in this list with AFib notifications, and they are free. WHOOP MG adds AFib alerts, an ECG and blood-pressure insights, but only in the WHOOP Life tier. Oura 's Blood Pressure Signals is a nighttime trend, not a cuff reading or a medical measurement. Ultrahuman 's AFib feature exists outside the US but is not available to US buyers. Polar has no built-in cycle tracking, and Amazfit does not document one. Data openness This is the dimension other comparisons skip. If you want your data in another app, a spreadsheet or an AI assistant, it matters who can read it and on what terms. Device Developer API Apple Health Health Connect WHOOP 5.0 Self-serve for 10 members; review above that Yes Yes Garmin CIRQA Application and approval One-way, from Garmin Connect One-way write Oura Ring 5 OAuth only; 10 users per app before approval; no data without a membership Yes; HRV, temperature, SpO2 and resting HR are not written Yes, including HRV Polar Loop Self-serve Yes, through Polar Flow Yes, through Polar Flow Google Fitbit Air Restricted scopes; verification and a yearly security assessment Yes, two-way; no HRV Yes Ultrahuman Ring PRO Self-serve personal token; apps through a portal Yes Yes Amazfit Helio Strap Partner-gated: enterprise verification Yes, through the Zepp app Yes, through the Zepp app Polar is the most open: a self-serve developer API, both phone health platforms, and no subscription in the way. Ultrahuman comes next, with a personal token any user can create and its own MCP server. WHOOP lets a small app start on its own and reviews anything larger. Garmin and Amazfit require a business application before an app can read your data. Oura fell furthest this year. It ended personal access tokens, and a Gen3 or later ring returns no data through the API without an active membership ( source ). Google retired the Fitbit Web API in autumn 2026. Its successor, the Google Health API, puts health data behind restricted scopes, app verification and a yearly security assessment ( source ). Value over three years Device Hardware Required subscription Optional subscription 3-year cost WHOOP 5.0 Included $199 to $359/year $597 to $1,077 Garmin CIRQA $199.99 None Connect+ $199.99 Oura Ring 5 $399 to $499 $69.99/year $608.97 to $708.97 Polar Loop $199.99 None $199.99 Google Fitbit Air $99.99 None Google Health Premium $99.99 Ultrahuman Ring PRO $479 None Ultrahuman Max $479 Amazfit Helio Strap $99.99 None Zepp Aura $99.99 The WHOOP membership includes the hardware and comes in three tiers: One at $199/year, Peak at $239/year and Life at $359/year ( source ). Over three years it is the most expensive device here. Oura is close behind once the membership is added. Every other device is a one-time purchase, and each optional subscription adds coaching or AI features, not your core data. Also considered Ultrahuman Ring AIR. Still sold outside the US. Ultrahuman does not ship it to US addresses. Samsung Galaxy Ring. Works with Android phones only. RingConn Gen 3. A budget ring with up to 14 days per charge and no subscription. Garmin Index Sleep Monitor. A screenless upper-arm band you wear only at night. The natural pick if you already wear a Garmin watch by day. Amazfit Helio Ring. Not on Amazfit's US store. Which one to buy You train seriously and want one device for recovery and workouts: Garmin CIRQA. Pick WHOOP 5.0 instead if recovery coaching matters more to you than the subscription cost. You care most about sleep and hate wearing anything on your wrist: Oura Ring 5. You refuse to pay a subscription: Polar Loop, Garmin CIRQA or Amazfit Helio Strap. You want your data to stay easy to move: Polar Loop. You want the lowest price with AFib alerts: Google Fitbit Air. You hate charging: WHOOP 5.0 or Ultrahuman Ring PRO. You lift and want reps counted: Garmin CIRQA, or Amazfit Helio Strap on a budget. Whichever one you choose, its data can do more than fill a chart in one app. freddy connects sources like Polar , Garmin and Oura to the AI assistant you already use, so you can ask about your sleep, training and recovery in plain words. Pricing and feature details verified 2026-10-30. Check vendor sites for current details before committing.

## Best wearable MCP servers in 2026: 8 options compared
URL: https://freddy.coach/articles/best-wearable-mcp-servers
Eight ways to connect your wearables to Claude, ChatGPT and other AI apps over MCP, compared on sources, setup and price. Updated 27 Oct 2026.
Your wearable knows how you slept. Your AI does not, unless something connects the two. That something is an MCP server, and in 2026 there are plenty to choose from. We compared eight options against each product's own pages on 27 Oct 2026, and added two built-in choices that need no MCP at all. freddy is ours, and we checked it the same way as the rest. This is not a ranking. The right pick depends on your phone, your devices and how much setup you want. At a glance Wearable MCP options at a glance (US dollars, checked 27 Oct 2026) Name Link Main pros Main cons Price freddy freddy.coach 100+ sources led by Garmin, Apple Health and Health Connect. Hosted, nothing to install. Works in Claude, ChatGPT and any MCP client. No coaches, dashboard or logging of its own. Free plan is 1 source and 7 days of history shown. Free; Pro $49 a year; Believer $249 lifetime Wellness Project wellnessproject.ai App with 8 AI coaches, logging and a dashboard, plus a hosted MCP server. Direct connections for Garmin, Fitbit, WHOOP, Withings, Polar and more. Free plan allows 3 coach chats a day, counted across the app and your AI. You work inside its own app. Free; Pro $9.99 a month Fulcra fulcradynamics.com MCP access on the free plan. One timeline for Apple Health, calendar and location, with an audit log of what agents read. Data comes in through its iPhone app. A general personal-context tool, so health is one part of it. Free; Pro $14.99 a month or $149.99 a year Health Auto Export healthyapps.dev iPhone app with a built-in MCP server. No account, and a low one-time or yearly price. Apple Health only. Your computer and phone must share a Wi-Fi network. The MCP server needs Premium after a 7-day trial. Basic $2.99 once; Premium $1.99 a month, $6.99 a year or $24.99 lifetime Garmin community MCP server github.com/Taxuspt/garmin_mcp Free, open source (MIT), runs on your computer. 110+ tools for activities, sleep, training load and more. Garmin only. You set it up and keep it running yourself. Free Open Wearables openwearables.io Open source (MIT) and self-hosted. The MCP server is in the free Core plan, and you keep all the data. You run it and get each provider's developer credentials yourself. Built for teams making products. Core free; Pro from $999 a month per deployment Terra AI tryterra.co One API for thousands of wearables, with an MCP interface for developers. Needs a developer account and API keys. Priced for companies, not individuals. From $399 a month billed annually, or $499 billed monthly Sahha sahha.ai Developer platform with an MCP server, plus ready-made scores, biomarkers and insights. 30-day free trial. For companies adding health data to an app. Not a personal tool. Starter $299 a month billed annually; Pro $1,499 a month billed annually Prices are in US dollars, taken from each product's own page on 27 Oct 2026. Check the link before you buy. What an MCP server actually does An MCP server is a small program that translates between an AI client and a data source. The client (Claude, ChatGPT, Cursor) speaks MCP. The data source (a wearable's cloud, your local files, a database) does not. The server bridges the two by exposing tools the AI can call, such as list_metrics , query_metrics and get_workout_summary . When you ask a question that touches your data, the AI calls those tools and gets back answers it can reason about. Personal health suits MCP well because the data lives in many places (watches, rings, apps, glucose monitors) and the questions often span several of them. A good health MCP server pulls from more than one source, puts the data in one place, and lets your AI do the reasoning. Hosted options for your own data 1. freddy What it is. A hosted MCP server from us. You connect your sources once, then add freddy to your AI app from Claude's or ChatGPT's directory, or paste your private MCP URL into any MCP client. Setup takes about 90 seconds. Sources. 100+, led by Garmin, Apple Health and Health Connect. Price. Free: 1 source and 7 days of history shown. Pro: $49 a year, with unlimited sources and full history. Believer: $249 lifetime. Good. One connection covers many devices. It is read-only, you can export your history as CSV, and deleting your account deletes your data. Not so good. freddy has no coaches, dashboard or logging of its own. It hands your data to the AI you already use. The free plan is one source with 7 days of history shown. Best for: people who wear more than one device, or anyone on Garmin, Apple Health or Health Connect who wants one connection across Claude, ChatGPT and other AI apps. 2. Wellness Project What it is. An app with eight AI coaches, a dashboard and logging for workouts, meals and sleep, plus a hosted MCP server. Claude, ChatGPT, Gemini, Grok and Mistral Le Chat can read the same data, and log new entries back. Sources. Direct connections for Garmin, Fitbit, WHOOP, Withings, Polar and more, plus Apple Health and Android Health Connect. Price. Free Basic, or Pro at $9.99 a month. Wearable connections are on both plans. Good. It is a full coaching app, not only a connector. Logging, coaches and a dashboard come with it. Not so good. Free Basic allows 3 coach chats a day, counted across the app and your AI connector. You work inside Wellness Project's own app and data store. Best for: people who want an all-in-one coaching and logging app, with an AI connector as a bonus. 3. Fulcra What it is. A hosted store for your personal context, with an MCP server for Claude, ChatGPT and other agents. It keeps Apple Health, calendar and location on one timeline, and logs what each agent reads. Sources. Apple Health, calendar and location through its iPhone app. Price. Free Personal plan with MCP access, 5 GB of files and 10M records. Pro is $14.99 a month or $149.99 a year. Good. MCP access is free, and you can share slices of your data with a coach or doctor. Not so good. Your data arrives through the iPhone app. Health is one part of a wider personal-context tool, so it is less focused on wearables. Best for: iPhone users who want one private timeline of health, calendar and location for their agents. Local and open-source options 4. Health Auto Export What it is. An iPhone app that exports your Apple Health data and includes a built-in MCP server, so an AI app on your computer can query your phone over your own network. Sources. Apple Health, so anything that writes to it. 150+ health and fitness metrics. Price. Free to look around, with a 7-day trial of Premium. Basic is $2.99 once. Premium is $1.99 a month, $6.99 a year or $24.99 lifetime, and the MCP server needs Premium after the trial. Good. No account, and the developer says it collects no data. The prices are low. Not so good. Your phone and computer must be on the same Wi-Fi network, and the server has no sign-in, so use it on a network you trust. Apple Health only. Best for: iPhone users who like to tinker and want the cheapest local setup. 5. Garmin community MCP server What it is. A free, open-source MCP server (MIT license) from the community, not from a company. It runs on your own computer and gives your AI 110+ tools for Garmin Connect data such as activities, sleep, training load and workouts. Sources. Garmin Connect only. Price. Free. Good. Free, deep Garmin coverage, and it runs on your own computer. Not so good. You install it and keep it running yourself, and it covers Garmin only. Best for: Garmin owners who are comfortable with a terminal and only need Garmin. 6. Open Wearables What it is. Open-source (MIT) infrastructure for wearable data that you run in your own cloud. Since our May version, it has shipped an MCP server in the free Core plan. Sources. Garmin, WHOOP, Polar, Suunto, Withings, Fitbit, Strava, Apple Health, Health Connect and more. You get each provider's developer credentials yourself. Price. Core is free forever. Pro starts at $999 a month per deployment. Good. You own the data and the stack, and there is no per-user fee. Not so good. It is built for teams making products. You run the servers, get the provider credentials and keep it working. Best for: developers and teams who need to keep wearable data in their own cloud. Built for developers and companies 7. Terra AI What it is. A wearable-data API used by app developers. Its AI interface exposes your users' connected data to AI tools over MCP. Sources. Thousands of wearables through one API. Price. From $399 a month billed annually, or $499 billed monthly, with 100,000 API credits included. Good. Broad device coverage in one API, and a 30-day risk-free trial. Not so good. It needs a developer account and API keys, and the pricing is for companies. Best for: teams adding wearable data to their own app. 8. Sahha What it is. A health-data platform for businesses, with scores, biomarkers and archetypes, and an MCP server over that data. Price. Starter is $299 a month billed annually. Pro is $1,499 a month billed annually. Every plan starts with a 30-day free trial. Good. Ready-made scores and insights, with a SOC 2 Type II report listed on every plan. Not so good. It is built to be added to an app, so it is not for individuals. Best for: companies that want scores and insights without building them. Built-in options, no MCP needed ChatGPT Health. Free, Go, Plus and Pro users in the US can connect Apple Health to ChatGPT on iPhone. Garmin, WHOOP and Oura data reaches ChatGPT if you let those apps share with Apple Health. It works only inside ChatGPT, and some app-specific scores may not carry over. OpenAI's help page has the details. Claude on iPhone. In beta for Pro and Max users in the US, Claude can read your Apple Health data from the iPhone app. It reads only and cannot write back. Claude's help page has the details. Microsoft lists Copilot Health as temporarily unavailable, so we left it out of the main list. Check Microsoft's support page for its status. Who each is best for Several devices, one connection: freddy. An all-in-one coaching and logging app: Wellness Project. A private timeline of health, calendar and location on iPhone: Fulcra. Cheapest local setup on iPhone: Health Auto Export. Garmin only, free, and you like the terminal: the Garmin community MCP server. Data in your own cloud: Open Wearables. Building an app for many users: Terra AI or Sahha. Just Apple Health and one AI app in the US: ChatGPT Health or Claude on iPhone. How to choose Start with your phone and your devices. If you use Apple Health only and one AI app, a built-in option may be enough. If you mix devices, or you want your data in whichever AI you use next, a hosted server saves the most effort. If you want to run everything yourself, pick one of the open-source options. If you are building a product for other people, look at the developer platforms. Your AI is only as useful as the data it can see, so the real question is which devices you wear and where you want their data to live. What changed since May 2026. We removed the separate Apple HealthKit and Google Health Connect entries, because they are developer frameworks and not servers (freddy still reads both). We moved Copilot Health out of the list while Microsoft has it paused. We added Wellness Project, Health Auto Export, a Garmin community server, Terra AI and Sahha. Open Wearables now ships its own MCP server, and Fulcra now has a free plan. Pricing and feature details verified 27 Oct 2026 from each product's own pages. Things move fast in this space, so check the vendor's site before you commit.

## Garmin data access for AI in 2026: what still works after the API changes
URL: https://freddy.coach/articles/garmin-data-access-ai
What still works for getting your Garmin data into an AI in 2026, and where freddy stands.
Ways to get your Garmin data into an AI (checked 26 October 2026) Name Link Main pros Main cons Price freddy https://freddy.coach/guides/garmin You sign in on Garmin's own screen and never share your password. One account for Garmin, Apple Health, Health Connect and 100+ sources. Works on web, desktop and phone. Free plan shows 1 source and 7 days of history. Garmin gives a new connection about a month of history; older history comes from Garmin's export. Free; Pro $49 a year; Believer $249 lifetime Garmin Personal API key https://developer.garmin.com/gc-developer-program/personal/personal-faq/ Official and free. No password goes into a script. Your own account only, pull only, and you need an MCP server on your own computer. Keys expire. Free Garmin Connect to Apple Health or Health Connect https://support.garmin.com/en-US/?faq=lK5FPB9iPF5PXFkIpFlFPA&productID=125677&tab=topics A setting in the Garmin Connect app. No developer account. Apple Health also feeds ChatGPT Health and Claude in the US. Shares only part of your Garmin data, one way. Health Connect needs Android 14 or newer. Free; your AI app has its own plan Garmin data export https://freddy.coach/guides/garmin Your complete archive, straight from Garmin. Manual, and a snapshot: nothing updates afterwards. Free Unofficial libraries such as python-garminconnect https://github.com/cyberjunky/python-garminconnect Reach nearly everything, with years of history. Your Garmin email and password go into a script, and the sign-in can stop working at any time. Free Garmin changed how apps get your data in 2026, and a lot of older advice no longer holds. This guide covers what changed, what still works for getting your own Garmin data into Claude, ChatGPT or another AI app, and where freddy stands. It is current as of 26 October 2026. freddy is ours. We say where it fits and where it does not. What changed in 2026 Garmin added a free personal key on 21 October. What still works 1. A service that connects through Garmin's official program You approve access on Garmin's own screen, and the service never sees your password. freddy works this way. So do a few other hosted connectors, which we list in our Garmin MCP servers roundup . This is the least work, and it works on your phone. 2. Garmin's Personal API key Launched on 21 October. In Garmin Connect on the web, open Settings, choose Personal API Key and create a key. Keys expire, so you will need to renew it. It covers your own account only. It is pull only: your code asks Garmin for data, and nothing arrives by itself after a workout. Garmin offers no MCP server for it, so you need one on your own computer. Community servers are starting to appear. 3. Apple Health and Health Connect The Garmin Connect app can share part of your data with Apple Health on iPhone, and with Health Connect on Android 14 or newer. Both are one way: Garmin writes to them and does not read back. On Android you have to switch it on yourself. Garmin shares the main activity, sleep and body metrics with Health Connect. Garmin-only metrics such as Body Battery may not come through. Once the data is in Apple Health, some AI apps can read it directly. ChatGPT Health connects to Apple Health for US users on iPhone. Claude's Apple Health integration is a US beta for Pro and Max. freddy also reads Apple Health and Health Connect, next to Garmin itself. 4. Your Garmin data export Garmin lets you request a full copy of your account data. It is a snapshot, so nothing updates afterwards, but it holds everything you have ever recorded. You can import it into freddy once to fill in older history. Our Garmin guide has the steps. 5. Unofficial libraries python-garminconnect and the MCP servers built on it sign in as you, with your Garmin email and password. They reach almost everything. The trade: your credentials sit in a script, and the sign-in can stop working at any time. If a Garmin tool asks for your Garmin email and password on its own page instead of sending you to a Garmin sign-in page, it uses the unofficial way. On your own computer, that can be a fair trade. Think twice before you type those credentials into a website somebody else runs. Where freddy stands freddy connects to Garmin through the Garmin Connect Developer Program. You approve access on Garmin's own screen, and freddy never sees your Garmin password. Your data is yours: when you delete your account, we delete your data. freddy reads Garmin, Apple Health, Health Connect and 100+ sources in one account. The free plan covers 1 source and shows 7 days of history. Pro is $49 a year and Believer is $249 once, for life. Paid plans have no source limit. See pricing . To add freddy, use the Claude connector directory or the official ChatGPT plugin . In Cursor or any other MCP client, you paste one personal URL. How to choose You want the least work, or more than Garmin in one place: freddy. You like to build your own and only need your own data: the Personal API key. You want something quick on your phone and partial data is fine: Apple Health or Health Connect. You want an archive: the Garmin data export. You are happy to trade a password for depth, on your own computer: an unofficial library. Last updated 26 October 2026. For other wearables, see our wearable MCP servers roundup .

## WHOOP MCP servers: 10 ways to get your WHOOP data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/whoop-mcp-servers
Ten ways to connect WHOOP to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with a TL;DR table, prices, and who each one suits.
TL;DR: WHOOP MCP options at a glance (prices checked on 26 October 2026) Name Link Main pros Main cons Price freddy freddy.coach WHOOP next to Garmin, Apple Health, and 100+ other sources. Nothing to install. WHOOP is on a waitlist right now. Free (1 source, 7 days of history shown); $49/year MissingMCP missingmcp.com Free, and one URL pasted into Claude. Claude only. Free AthleteData athletedata.health Hosted, 30+ apps, optional coach. Paid from day 8. Not a WHOOP-only tool. $9/mo or $69/yr Wellness Project wellnessproject.ai Coaching app with a free plan. Free plan allows 3 analysis chats a day. App first, AI second. Free; Pro $9.99/mo usefulapi usefulapi.io Hosted URL that works in Claude, ChatGPT, and Cursor. Its login asks for a WHOOP API key. Very small free allowance. Free tier; Pro $9/mo or $90/yr FitnessPeak fitnesspeak.app Hosted and built around WHOOP. No developer account. Card needed for the trial. ChatGPT's connector screen does not work with it. $39/yr @nchemb/whoop-mcp GitHub Free, local copy of your data, no developer app. Needs Node 22. Desktop apps only. Free whoop-ai-mcp GitHub Most-used open-source option, under your own WHOOP app. You register a developer app. Desktop apps only. Free totem GitHub Deepest tool set, writes included. Uses WHOOP's private API. Risk to your account. Free; cloud host may charge Open Wearables GitHub Many wearables in one self-hosted server. The most setup, and you run the server. Free; you pay for hosting WHOOP tracks your recovery, strain, and sleep around the clock, and the AI assistant you talk to every day cannot see any of it. The Model Context Protocol ( MCP ) is the open standard that connects the two, and ten options now do it for WHOOP. They differ in how they reach your data, and in what they ask of your WHOOP account. This list ranks them by setup effort, from least to most. freddy is ours, and it is first: nothing to install, no password shared, and WHOOP next to your other apps in any AI app that speaks MCP. WHOOP is on a waitlist in freddy right now, so we say what that means below. Each entry says who it suits. What to know first WHOOP has an official developer API. Every official route here connects through WHOOP's own sign-in screen. There is also an unofficial route: the private API behind WHOOP's iPhone app. It reaches more, writes included, but it needs your WHOOP password and a text-message code if your account uses one. One project on this list uses it, and it carries a warning. A tool that asks for your WHOOP email and password uses WHOOP's private API. It can stop working at any time. If your WHOOP account matters to you, stay with the official route. Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. The list 1. freddy freddy is a hosted MCP server that brings your health data from all your apps into one place. It connects Garmin directly, reads Apple Health and Health Connect, and covers 100+ sources. You add freddy to your AI app from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. WHOOP is on a waitlist. Join it from your dashboard and connect your other sources in the meantime. Once a spot opens, you connect WHOOP on WHOOP's own sign-in screen, so freddy never sees your password. Setup: join the waitlist, then three steps in your browser once your spot opens. Nothing to install, and it works on your phone too. Price: free for 1 source and 7 days of history shown, or $49/year for all your sources and your full history. Paid plans have no source limit. Best for: people who use more than one app and would rather not run a server or share a password, and who can wait for a WHOOP spot. 2. MissingMCP MissingMCP is a free hosted gateway run by one developer. For WHOOP it uses the official API through its own WHOOP app, with 8 read-only tools, and you approve access on WHOOP's screen. Its WHOOP page covers Claude only. Setup: paste one URL into Claude and approve access on WHOOP's screen. Price: free. The operator accepts donations. Best for: Claude users who want WHOOP in their AI app today, at no cost. 3. AthleteData AthleteData is a hosted AI coach with a cheaper MCP-only plan. It connects more than 30 apps, WHOOP among them, through its own WHOOP connection, and keeps your WHOOP data up to date from WHOOP's update notifications. The coaching plan messages you after workouts; the MCP-only plan leaves that out. Setup: start the trial, connect WHOOP, add the connector address to your AI app, and sign in. Price: MCP-only is $9 a month or $69 a year, after a 7-day free trial ( plans ). Coaching is $39 a month. Best for: athletes who want a training coach as well as answers. 4. Wellness Project Wellness Project is a hosted coaching app with its own MCP server. It links WHOOP inside its app, keeps the synced history, and offers it to Claude and ChatGPT. The vendor lists 73 tools, and the ones that write change Wellness Project, not WHOOP. The free plan allows 3 analysis chats a day, with the app and your AI app counted together. Setup: install the app, connect WHOOP under Settings and Devices, then add its connector to your AI app and sign in. Price: a free Basic plan. Pro is $9.99 a month, and local prices may vary ( pricing ). Best for: people who want a coaching app alongside their AI assistant. 5. usefulapi usefulapi is an independent hosted service with a WHOOP MCP server. Its tools are read-only and cover recovery, sleep, cycles, workouts, and body measurements. It lists Claude, ChatGPT, Cursor, and VS Code as clients. One thing to check first: its login asks for a WHOOP API key rather than sending you to WHOOP's own sign-in screen. Read its page on how that works before you connect. Setup: paste its URL into your AI app as a custom connector, then enter your WHOOP API key when asked. Price: a free tier with a very small monthly allowance, then Pro at $9 a month or $90 a year ( pricing ). Best for: people who want a hosted URL in several AI apps and are fine creating a key. 6. FitnessPeak FitnessPeak runs a hosted MCP server built around WHOOP, with 5 read-only WHOOP tools and a status tool. You connect WHOOP once on its site, approve on WHOOP's own screen, and need no WHOOP developer account. It connects with a key instead of a sign-in screen, so it suits Claude, Cursor, VS Code, and other clients that accept a key. Its own FAQ says ChatGPT's connector screen does not work with it. Setup: sign up with a card, connect WHOOP, copy your key, and paste its URL and key into your AI app. Price: $39 a year, after a 14-day trial ( pricing ). Best for: people who want a small hosted server built around WHOOP, without a developer account. 7. @nchemb/whoop-mcp A local server that needs no WHOOP developer app: 6 tools, including read-only SQL over a local SQLite copy of your data, under an MIT license. It signs you in through the author's shared WHOOP app. Your data stays on your computer, but each token refresh passes through the author's sign-in service. Setup: moderate. Install Node 22.13 or later, add it to Claude Code with one command, and run its init command to sign in. Desktop apps only. Price: free and open source. Best for: people who want a local copy of their data without registering a developer app. 8. whoop-ai-mcp The most-used open-source WHOOP server, with 175 stars, about 2,700 npm downloads a month, an MIT license, and 16 read-only tools on the official API. Its repository says it is not affiliated with WHOOP. You register your own WHOOP developer app, so the app is yours alone. Setup: involved. Install Node 20 or later, create a WHOOP developer app with a local redirect address, run its setup command for your AI app, and approve access in the browser. Desktop apps only, unless you host it yourself. Price: free and open source. Best for: developers who want the official API under their own app. 9. totem The deepest WHOOP server: 55 tools , 15 of which write, built on the private API behind WHOOP's iPhone app. It signs in with your WHOOP email, password, and a text-message code if your account uses one. It can deploy to a cloud host such as Fly, Railway, or Cloud Run, so it can work in Claude on the web and phone, and in ChatGPT. The risk is your WHOOP account. Setup: involved. Install Node 24, clone and build it, sign in, and deploy it with its cloud command. Sign in again about every 30 days. Price: free and open source. The cloud host may charge. Best for: people who want everything, writes included, and accept the risk to their account. 10. Open Wearables An open-source platform that you run on your own server, with 2,717 stars and an MIT license. It signs in to WHOOP with your own developer app, syncs your data into its own database, and includes an MCP server with 6 tools. It covers several other wearables too. Setup: the most work on this list. Run its Docker stack, create your own WHOOP developer app, put its keys in the config, create an API key, and set up the MCP server. Desktop apps only. Price: free software. You pay for the server. Best for: developers who want several sources on hardware they control. How to choose For most people, the answer is freddy: no install, no password, and WHOOP next to your other apps in any AI app that speaks MCP. WHOOP is on a waitlist right now, so join it from your dashboard. If you need WHOOP in your AI app today: You use Claude and want it free: MissingMCP. You want a coach, or several apps in one place now: AthleteData. You want a hosted URL in several AI apps: usefulapi or FitnessPeak. You want a local copy without a developer app: @nchemb/whoop-mcp. You want the official API under your own app: whoop-ai-mcp. You want writes and accept the risk to your account: totem. Prices are in US dollars and were checked on 26 October 2026, along with star counts and download numbers. Other third-party details were checked in September and October 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## What's new in freddy: week of Oct 19, 2026
URL: https://freddy.coach/articles/whats-new-2026-10-25
Week 43: the 22 Oct outage and what we found, complete food and drink logs, a direct URL without your key, big Garmin syncs and Health Connect sleep that adds up.
Good morning, health and performance enthusiasts. A warm welcome to the 1,025 new members who joined freddy last week. In short On 22 Oct freddy was unavailable for about 6 hours and 40 minutes. It is fixed and your data is safe. What happened Your direct URL no longer contains your key. Apps that can't sign in send your MCP key as a header instead, and your old direct URL keeps working for now. Food and drink entries from Apple Health and Health Connect logged at the same moment no longer overwrite each other. Health Connect sleep: a night no longer reads longer than it was when its sleep stages arrive later. Garmin: very large syncs, like long and detailed activities, no longer fail. We also squashed several bugs and errors. Use case of the week: how Nate brings his training context together in ChatGPT. What happened on 21 to 22 Oct From 23:22 UTC on 21 Oct to 06:03 UTC on 22 Oct, freddy was unavailable for about 6 hours and 40 minutes. It is fixed and your data is safe. We're sorry for the disruption. The full write-up is here: What happened . Your direct URL no longer contains your key Your direct URL is now tied to your freddy account and no longer has your key in it. An app or script that can't sign in sends your MCP key as a header instead. The URL needs that key, or a sign-in, to work. Your old direct URL keeps working for now. Find the new one on the Direct URL card in your dashboard. Every food and drink entry is kept If you log meals in an app that writes to Apple Health or Health Connect, entries logged at the same moment could overwrite each other, so some nutrition values went missing. Each entry is now stored separately, so new nutrition data arrives complete. Health Connect sleep adds up When Health Connect first sent a night without stages and later sent the staged version, the night could read longer than it actually was. New syncs now replace the old total with the stages. Big Garmin syncs go through A very large Garmin sync, such as a long activity with detailed data, could fail. freddy now accepts it, so nothing is left behind. Use case of the week: Nate's training context, together in ChatGPT Nate is building a private, iPhone-first AI fitness coaching system inside ChatGPT. It's a personal project in active development. freddy provides his workout performance, activity and training-load insights, which Nate combines with Motra, Peloton, Apple Health, Foodnoms, Google Calendar and Drive. The goal is coaching recommendations based on his actual performance and trends: strength progressions, recovery needs, training adjustments and nutrition guidance. He's also working toward approval-based actions that can make and verify changes across connected services. Ultimately, Nate wants to move beyond tracking fitness data to a coach that adapts and improves alongside him. Have a freddy use case you'd like to share? Send it to support@freddy.coach with the question you asked and what made the answer useful.

## Garmin MCP servers: 10 ways to get your Garmin data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/garmin-mcp-servers
Ten ways to connect Garmin Connect to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, plus how Garmin's new Personal API key fits in.
Garmin Connect holds years of your sleep, training load, Body Battery, and workouts, and none of it reaches the AI assistant you talk to every day. The Model Context Protocol (MCP) is the open standard that connects the two, and at least ten projects now do it for Garmin. They differ a lot in how much work they take. This list ranks them by effort, from least to most: what you install, what you configure, and what you have to keep running. freddy is ours, and it is first because it is the least work: nothing to install, no config file, no password shared, and one account for Garmin and your other apps. Each entry says who it suits. Two things to know first Garmin has two ways in, and every tool below uses one of them. The choice decides how much history you get, whether you type your password into the tool, and how likely the tool is to break. The official way is the Garmin Connect Developer Program . You approve access on Garmin's own screen, and the tool never sees your password. Garmin also added a personal key for individuals, covered in the next section. A new connection reaches back 31 days, and everything after that arrives as you train. The unofficial way is python-garminconnect , a community library that signs in as you, with your email and password. It reaches everything, years of activities included. The cost: your credentials go into a script, and Garmin can change its private sign-in at any time. It did in 2026, and garth , a popular Garmin login library, was retired as a result. If a Garmin tool asks for your Garmin email and password instead of sending you to a Garmin sign-in page, it uses the unofficial way. On your own computer, that can be a fair trade. Think twice before you type those credentials into a website somebody else runs. The second thing is where the server runs. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server that runs on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. New in October 2026: Garmin's own Personal API Until now, an individual had no official way to get programmatic access to their own Garmin data. On October 21, 2026, Garmin launched one: the Garmin Connect Personal API . It is free, and you set it up yourself in a few clicks. How it works. Sign in to Garmin Connect on the web, open Settings, choose Personal API Key and create a key. Keys expire, so you will need to renew it. What you get. Activity data, including .FIT files, plus daily health data. Delivery is pull only: your code asks Garmin for data, and nothing arrives by itself after a workout. You cannot send workouts or routes to your watch with it. What to know before you rely on it: It covers your own account only. Garmin says it is not for commercial apps that serve several users , so it is a way to build something for yourself, not a way for a service to connect you. Garmin does not offer an MCP server for it, so you need one in front of it, and it has to run on your own computer. Community servers built on the Personal API are starting to appear. Some advanced Garmin fields may not come through the Personal API. The Personal API can return your full history. Keys expire, so plan to renew one at least once a year. Where it fits. The key removes the biggest problem with the unofficial libraries below: no password goes into a script. If you like to tinker and want your own data on your own machine, it is a good option. If you would rather skip the setup, freddy does the same job in a few minutes: you connect Garmin on Garmin's own screen, add freddy from the Claude connector directory or the official ChatGPT plugin , and it works on your phone too, with your other apps in the same place. The list 1. freddy freddy is a hosted MCP server for your health data, with Garmin as one of its sources. You sign in with a code we email you, connect Garmin on Garmin's own sign-in screen, and add freddy to your AI app. In Claude, connect from the Claude connector directory . In ChatGPT, use the official freddy plugin . In Cursor or any other MCP client, you paste one personal URL. freddy connects to Garmin through its official developer program, so it never sees your Garmin password. New Garmin data arrives by push, so a run is ready to ask about soon after it syncs. The same account also reads Polar, Withings, Suunto, Hevy, Intervals.icu, Concept2, Wahoo, Ultrahuman, Oura, Apple Health, and Health Connect, so a question like "did I sleep worse in the weeks I lifted heavy" works across all of them. The limit is Garmin's 31-day reach for a new connection. For older history, request your data export from Garmin and import the ZIP into freddy once. Our Garmin guide walks through every step. Setup: three steps in your browser. Nothing to install, nothing to update, and it works on your phone too. Price: free for one source with a short history window, or $49/year for all your sources and your full history. Best for: almost everyone, and especially people who use more than one app or would rather not run a server or share a password. 2. Fitness AI Connector A hosted MCP server for Garmin alone , built on the official Garmin API. You sign in through Garmin, then add it to ChatGPT from the Plugin Directory or to Claude as a connector. It has 12 tools for daily health, activities with lap detail, body composition, and cycle tracking, and the project reports more than 40,000 users. It covers Garmin only, and your history starts on the day you connect. Setup: a few steps in your browser, nothing to install. Price: a free tier, and a Basic plan at $3 a month. Best for: Garmin-only users who want a hosted server at the lowest price. 3. AthleteData A hosted AI coach that gets in touch when your training needs a change and sends the next session to your watch. It connects more than 30 apps, Garmin among them, through the official API, with about a month of starting history and an export upload for more. Coaching is the main product. The plan that matters here is MCP-only: you query your data from Claude, ChatGPT, or Cursor, without the coach. Setup: connect your apps, copy an API key from the dashboard, and add the key and server address to your AI client. Price: MCP-only is CHF 9 a month or CHF 69 a year, after a 7-day free trial ( pricing ). Coaching is CHF 29.90 a month. Best for: people who want a coach that gets in touch first, not only one that answers. 4. Tredict An endurance training platform with its own MCP server and ChatGPT plugin. You connect Garmin through Garmin's sign-in page, and Tredict can import your last two years of activities. From the chat, it can build a training plan and send the workouts to your watch. It works best if Tredict is already your training log. The AI sees Tredict's copy of your data, with workouts at the center and HRV, sleep, and weight alongside. Setup: create a Tredict account, connect Garmin, then add Tredict to your AI client. Price: 2 months free, then $49 for 12 months, paid once (€49 in Europe; prices ). Best for: endurance athletes who already plan their training in Tredict. 5. MissingMCP A free hosted gateway that runs entry 7's open-source server for you. Your AI client connects to the gateway, and the gateway's sign-in page asks for your Garmin email and password. The operator says the password is not kept, only encrypted session tokens. The gateway code and the upstream server are both public. You get the depth of the unofficial way, with your whole history and workout scheduling, and nothing to install. Your credentials pass through a service one person runs, which is a different risk from a company in Garmin's program or a script on your own computer. Setup: paste a URL into your AI client and sign in with your Garmin account. Price: free. The operator accepts donations. Best for: people who want the unofficial depth without a terminal, and accept the password trade. 6. eddmann/garmin-connect-mcp The compact local option: 22 tools in eight groups plus six ready-made prompts, MIT licensed. Fewer tools can help: a model chooses among 22 clear tools faster than among 110. Setup: moderate, on your own computer. Install uv or Docker, run a sign-in command in a terminal with your Garmin email, password, and MFA code, then add a block of JSON to your AI app's config file. Desktop apps only. Price: free and open source. Best for: people at home in a terminal who want a small, tidy local server. 7. Taxuspt/garmin_mcp The original community Garmin MCP server, and the base for many others: 1,225 stars , MIT licensed, and updated this month. It is the deepest option here, with more than 110 tools: activity edits, workouts it can build and schedule onto your Garmin calendar, gear, nutrition logs, FIT files, and training-load trends. Your full history comes with it. Setup: moderate to involved. Claude Desktop gets a one-click extension, but you still sign in once from a terminal with Python 3.12 and your Garmin password, and again when the tokens run out, about every six months. Other apps need a config file edit. Desktop apps only. Price: free and open source. Best for: technical Garmin users who want every field and keep their credentials on their own hardware. 8. Nicolasvegam/garmin-connect-mcp The TypeScript alternative: 61 tools that cover activities, sleep, daily health, body composition, trends, training, and devices. Pick it over entry 7 if you would rather run Node than Python. The repository has no license file, so its terms are unclear. Setup: moderate. You need Node.js 20, your Garmin email and password in your AI app's config file, and a one-time terminal command if your account uses MFA. Desktop apps only. Price: free. Best for: Node developers who want a broad tool set. 9. GarminDB plus a SQLite MCP server Not an MCP server by itself. GarminDB (3,316 stars, GPL-2.0) downloads your Garmin Connect data into SQLite databases on your disk, with the username and password in its config file. A generic SQLite MCP server then lets your AI app query those files. You own the archive: every activity you ever recorded, in a format that outlives every vendor on this list. Your questions become SQL, which suits trends and totals better than "what did I do yesterday". Setup: involved. Install GarminDB with Python, edit its config file, run the first download (long, for years of data), set up a separate SQLite MCP server, and re-run the sync whenever you want fresh data. Price: free. Best for: people who want the archive first and the AI second. 10. Open Wearables An open-source platform that you run on your own server, with 2,550 stars and an MIT license. It puts several providers behind one API and includes its own MCP server. It covers Garmin, Polar, Suunto, WHOOP, Fitbit, Oura, Withings, and Ultrahuman, plus Apple Health, Health Connect, and Samsung Health through mobile SDKs. It is the one way here to get several sources on hardware you control. Setup: the most work on this list, with a catch for Garmin. You run it with Docker on a server and connect each provider with your own developer keys. Garmin's business keys need an approved application. Price: free software. You pay for the server. Best for: self-hosters who want full control of their other sources, and businesses that already hold Garmin keys. Garmin MCP options at a glance (prices as of September 2026) Tool Setup Garmin sign-in History Price freddy None; web, desktop, and phone Official 31 days, then forward; export import Free tier; $49/year Fitness AI Connector None; web, desktop, and phone Official From the day you connect Free tier; $3/mo AthleteData API key; web, desktop, and phone Official About 30 days; export upload CHF 69/yr or CHF 9/mo Tredict Tredict account; web, desktop, and phone Official Last 2 years 2 months free, then $49 per 12 months MissingMCP None; web, desktop, and phone Email and password Full Free eddmann Terminal and config file; desktop only Email and password Full Free Taxuspt/garmin_mcp Terminal and config file; desktop only Email and password Full Free Nicolasvegam Node and config file; desktop only Email and password Full Free GarminDB + SQLite MCP Two tools and a re-sync; desktop only Email and password Full Free Open Wearables Your own server and Garmin approval Your own Garmin business key Depends on your key Free; you pay for hosting Garmin Personal API + your own MCP server Key from Garmin Connect, plus a server you run; desktop only Your own Garmin key (no password) Back to first data in our test Free How to choose For most people, the answer is freddy. It is the least work, it never sees your Garmin password, and it works in any AI app that speaks MCP, on your phone too. The exceptions: Garmin is your only app and price comes first: Fitness AI Connector. You want a coach that gets in touch first: AthleteData. You already plan your training in Tredict: its own connector. You live in a terminal and want every Garmin field: Taxuspt/garmin_mcp. You want everything on hardware you control: GarminDB for a Garmin archive, or Open Wearables if you already hold Garmin developer keys. You want to build your own on Garmin's official key, with no password in a script: the Personal API section above. Prices and details verified in September 2026. The Garmin Personal API section was added on October 24, 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Service incident: freddy outage on 22 October 2026
URL: https://freddy.coach/articles/incident-2026-10-22
freddy was unavailable on 22 October 2026. Service is restored and we are monitoring.
Status: Resolved. freddy is working normally and we are monitoring. Last updated 22 October 2026, 06:25 UTC. We'll keep this page updated as we learn more. What happened freddy was unavailable from 23:22 UTC on 21 October to 06:03 UTC on 22 October 2026. During this time the freddy website, the app and AI assistant connections were unavailable. freddy came back online at 06:03 UTC. Your data We've found no evidence that anyone accessed your data. Keeping your data safe is our highest priority, and we're taking further steps to keep an outage like this from affecting freddy again. What you need to do Nothing. Your connected sources catch up on their own. Data from recent days may take a while to fill in. Updates All times are in UTC, newest first. 06:25 UTC. Monitoring: We've checked our systems and the sync backlog has cleared. We're continuing to monitor. 06:03 UTC. Resolved: freddy is back online. The website, the app and AI assistant connections are working. Your sources are catching up on missed data, and recent days may take a while to fill in. 05:53 UTC. Recovering: We've emailed users about the outage. Recovery is in progress. 05:16 UTC. Recovering: Our restart went through. Recovery is in progress. 05:00 UTC. Investigating: freddy is still unavailable. We're still working on restoring service and continue to investigate. Next update in about 30 minutes. 04:30 UTC. Investigating: freddy is still unavailable. We're still working on restoring service and continue to investigate. Next update in about 30 minutes. 04:00 UTC. Investigating: freddy is still unavailable. We're still working on restoring service and continue to investigate. Next update in about 30 minutes. 03:30 UTC. Investigating: freddy is still unavailable. We're still working on restoring service and continue to investigate. Next update in about 30 minutes. 03:00 UTC. Investigating: freddy is still unavailable. We're still working on restoring service and continue to investigate. Next update in about 30 minutes. 02:30 UTC. Investigating: freddy is still unavailable. We're still working on restoring service and continue to investigate. Next update in about 30 minutes. 02:00 UTC. Investigating: We have no evidence that anyone accessed user data. 01:30 UTC. Investigating: freddy is still unavailable. We're still working on restoring service and continue to investigate. Next update in about 30 minutes. 01:00 UTC. Investigating: We've posted an update on X ( @hi_freddy_coach ) and added a status notice on freddy.coach. 00:49 UTC. Investigating: We are seeing an outage and are investigating. 23:22 UTC on 21 October. Outage: freddy went down. The website, the app and AI assistant connections are unavailable. Thank you for your patience. freddy team

## What's new in freddy: week of October 12, 2026
URL: https://freddy.coach/articles/whats-new-2026-10-18
Full Personal API records, daily summaries with your own range, automatic time zones, app logins that no longer expire, Withings sleep and Claude connector news.
Good morning, health and performance enthusiasts. A warm welcome to the 733 new members who joined freddy last week. Week 42 brings daily summaries, account time zones, sign-ins without a time limit, a Withings sleep fix, and full records on the Personal API. There is also recent Claude connector news from the previous week. In short Personal API: fetch the full record behind each metric, with the device and record id on every row. Daily summaries: ask your AI for daily averages, lows and highs of a metric. Automatic time zones: freddy follows your device's time zone, or you can pick one in Settings on the web. Your freddy app login no longer expires after 30 days, so your sync keeps running. Withings sleep: detailed sleep data now keeps updating, not just the summary. From last week: freddy is now an official Claude connector. Use case of the week: try a daily symptom scan to spot early signals that you might be run down. Personal API: the full record behind each metric The Personal API now returns the same detail your AI sees. Add include_raw=true to GET /api/v1/metrics to fetch the stored record behind each row, such as exercise sets, sleep stages and heart-rate streams. Every row also carries its timestamp, device and source record id, and you can filter by device or source id. The Personal API is in beta. Daily summaries: averages, lows, highs and your own range Ask your AI for per-day averages, lows and highs of a metric, plus the share of readings inside a range you set. For example, give a glucose range of 70 to 180 mg/dL to see how many readings fell below, inside and above it. Weekly and monthly summaries work the same way. Automatic time zones freddy now follows your device's time zone automatically, so your data is grouped into the right days. To keep a particular zone, pick one in Settings on the web. If you chose a time zone yourself, freddy asks before replacing it with a different detected zone. Your AI can see the account time zone and where it came from, so a daily view has that context. Your app login no longer expires Sign-ins used to end after 30 days, which signed you out of the app and stopped your sync until you signed in again. A sign-in now lasts until you sign out or revoke it in Connected devices, so your sync keeps running. On the web, signing in again in another tab also no longer lets an older request sign you out of the new session. Withings sleep: detailed data keeps updating Detailed Withings sleep data could stop updating after the first connection, even while sleep summaries kept arriving. New sleep updates now fetch both. This fix is live for existing connections. There is no new setting to enable. From last week: freddy is an official Claude connector freddy is now an official Claude connector. Connect your freddy account once, then ask Claude about your health and performance data in any conversation. If you already use freddy with another AI client, use the same freddy account. Your connected sources and stored health history stay with that account. Use case of the week: try a daily symptom scan Each morning, ask your AI to scan last night and the past few days of your wearable data for early signals that you might be getting sick or run down, then suggest whether to take it easy. Depending on your devices, freddy can share resting heart rate, HRV, sleep, skin or body temperature and respiratory rate. Comparing them with your own baseline shows what is unusual for you. Try: Use freddy to compare last night and the past three days with my 30-day baseline. Flag changes in resting heart rate, HRV, sleep, temperature and respiratory rate, note any missing data, and tell me whether to take it easy today. Have a freddy use case you'd like to share? Send your idea to support@freddy.coach with the question you asked and what made the answer useful. Leave out personal health details, and say whether to credit you or keep it anonymous.

## What's new in freddy: official Claude connector and steadier source updates
URL: https://freddy.coach/articles/whats-new-2026-10-11
freddy is now an official Claude connector, Intervals.icu updates are more reliable, and Ultrahuman sign-in works again.
A warm welcome to the 735 new members who joined freddy last week. Week 41 gives freddy an official home in Claude's connector directory. It also includes a reliability improvement for Intervals.icu and an Ultrahuman sign-in fix. In short freddy is now an official Claude connector. Find it in Claude's connector directory and add it directly. Intervals.icu updates now keep unfetched days available for a later sync. Catch-up: Ultrahuman sign-in works again with a fresh Access Code from the Ultrahuman app. Use case of the week: switch your AI while keeping your health history in freddy. Privacy notice: freddy plans to add OpenAI and Anysphere to its subprocessor list before either service receives personal data. freddy is now in Claude's connector directory You can add freddy from Claude's connector directory instead of setting it up as a custom connector. Connect your freddy account once, then use the same health and performance context in any Claude conversation. The connector remains permission-based. Claude receives data only when you connect your freddy account and use the connector. You can disconnect it again from Claude or freddy. More reliable Intervals.icu updates A small Intervals.icu update does not always need to fetch every kind of data. The update path now keeps days it did not fetch available for a later sync instead of treating them as already covered. The result is a steadier path for collecting anything that a narrow update did not request. There is nothing new to configure. Catch-up: Ultrahuman sign-in works again If the connection asks for a six-character code, create a fresh one in the Ultrahuman app under Profile, then Access Code. freddy now shows this path before you connect. Use case of the week Been using ChatGPT to review your training and want to try Claude? Add freddy in Claude and sign in to the same freddy account. Your sources stay connected, and Claude can query the history already stored in freddy. Your chat history does not transfer between AIs, so bring your goals, schedule and preferences into the new conversation. If you have workout, sleep or recovery data from the last seven days, try: Use freddy to pick up my training context. Summarize my workouts, sleep and recovery from the last seven days, and flag missing data. My goal is [goal], and I can train on [days]. Ask what else you need to help me plan the week ahead. Try the same question in both AIs, compare their approaches, or continue with the one you prefer. Your freddy setup stays in place. Subprocessor notice freddy plans to add OpenAI and Anysphere, the company behind Cursor and Grok Bot, to its subprocessor list. Before either service receives personal data, the privacy policy will state the purpose and the data involved. This notice does not change current processing and does not require any action today.

## What's new in freddy: week of September 28, 2026
URL: https://freddy.coach/articles/whats-new-2026-10-04
Date-based questions now follow your local day, and daily totals tell your AI when they may still be incomplete.
A warm welcome to the 736 new members who joined freddy last week. Week 40 makes two everyday questions more literal: which day did a reading belong to, and is a daily total actually finished? In short Date ranges now follow the local day reported by the freddy app, so morning workouts and other readings stay on the day you experienced them. Daily totals now tell your AI when they may still be incomplete and when that source was last read. Dates now match the day you experienced A date can mean different things depending on where you are. Early in the morning in Seoul, for example, the UTC calendar can still say the previous day. A workout or reading could therefore disappear from a date-based question even though it was clearly part of that day for you. Date-only questions now use the timezone reported by the freddy app. A request for Tuesday follows Tuesday where you are, while source-provided daily records keep the day assigned by the source. If freddy does not know your timezone, the answer says that it is using UTC instead of leaving the rule unstated. Daily totals no longer look more final than they are A total for today is a moving snapshot. The same can be true for an older day if a source last read it before that day ended. Until now, that value could look just as settled as a complete full-day result. freddy now adds a possibly incomplete note when the day is still open or a source was last read before it ended. The note includes the last read time. Your AI can then describe the total as a snapshot that may still grow, rather than treating it as the final word on the day. What changes in a conversation Questions such as "Show my workouts from Tuesday" now use the day you meant when the app has reported your timezone. Questions about today's steps, sleep, or other daily totals carry the extra context needed to distinguish an in-progress day from a finished one. The change applies inside freddy's AI tools, so it follows you across Claude, ChatGPT, Perplexity, and other compatible clients. There is nothing new to configure beyond keeping the timezone on your phone current.

## What's new in freddy: week of September 21, 2026
URL: https://freddy.coach/articles/whats-new-2026-09-27
Lyfta and Runalyze for everyone, a new way to connect multiple accounts and headless agents, stronger mobile sync recovery and easier account switching.
A warm welcome to the 708 new members who joined freddy last week. Here’s what changed: In short Lyfta and Runalyze are now open to everyone in freddy. A new (sorta) way to connect to freddy. Mobile 1.0.15 improves Apple Health and Health Connect sync recovery. AI reconnects now show the signed-in account and let you switch. Lyfta and Runalyze are open to everyone Lyfta and Runalyze have moved out of beta. You can now connect either source from the Sources page without joining a waitlist. This makes both connections part of the standard source catalog for every freddy account. Lyfta connects with an API key from Lyfta API Access. Runalyze connects with a personal API token. Runalyze requires a Supporter or Premium account because its API access comes from that service. A new (sorta) way to connect to freddy Direct connection links now make two things possible: connect multiple freddy accounts to the same AI client, and connect headless agents or other clients that cannot open a browser sign-in. Every account has a private link in Settings on the web and in the app. Reveal or copy it when needed, rotate it if it should stop working, and keep it private like a password. Mobile 1.0.15 improves health sync recovery Apple Health re-imports now show honest progress while recent data continues to move. The app keeps the phone awake during the foreground walk, which makes a long re-import less likely to stop before it finishes. Current days remain available while older history catches up. Health Connect now recovers better from interruptions and unreadable records. One record that Android cannot read no longer has to freeze the rest of a changes feed or prevent a re-import from finishing. Daily activity totals are also more consistent on supported Android versions. Update to freddy 1.0.15 from the App Store or Google Play to use these mobile changes. Reconnect with the account you want When you reconnect an AI app, freddy now shows which account is signed in before you continue. Choose Continue to use that account, or choose Use another account to switch first. This makes the account choice explicit before the AI app receives access. Signing out on the reconnect page now clears the server session. The next sign-in starts with the account you choose instead of silently returning to the previous session.

## Concept2 MCP servers: 8 ways to get your Concept2 Logbook into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/concept2-mcp-servers
Eight ways to connect your Concept2 Logbook to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with prices and where each one reads from.
Your Concept2 Logbook holds every row, ski, and bike session you have done on an erg, and the AI assistant you talk to every day cannot see any of it. The Model Context Protocol (MCP) is the open standard that connects the two, and eight options now do it for Concept2. This list ranks them by setup effort, from least to most. freddy is ours, and it is first: nothing to install, no token to paste, and your erg sessions next to your sleep and recovery in any AI app that speaks MCP. Each entry says who it suits. What to know first The question that sorts Concept2 options is where they read from. A tool that reads the Logbook directly sees every result, hand-entered ones included. A tool that reads a copy in another training platform sees only what that platform imported. The Logbook API is open to any Logbook user, free, with a sign-in for apps and a long-lived personal token for your own use, under Edit Profile, Applications ( Concept2 API docs ). The token opens your whole Logbook, so keep it in a config file on your own computer. Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. The list 1. freddy freddy is a hosted MCP server that brings your health data from all your apps into one place. You sign in with a code we email you, connect Concept2 on the Logbook's own sign-in screen, and add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. freddy reads your Logbook directly, read-only, so every result comes through, and new sessions arrive soon after you log them. The same account reads Garmin, Oura, Polar, Apple Health, and more, so a question like "do my 2k splits track my sleep" works across all of them. Our Concept2 guide walks through every step. Setup: three steps in your browser. Nothing to install, nothing to update, and it works on your phone too. Price: free for one source with a short history window, or $49/year for all your sources and your full history. Best for: almost everyone who rows, skis, or rides on a Concept2 erg. 2. AthleteData AthleteData is a hosted AI coach with a cheaper MCP-only plan. It signs in to your Logbook, reads it, and imports your history when you connect, next to more than 30 other apps. The coaching plan messages you after workouts; the MCP-only plan leaves that out. Setup: start the trial, connect Concept2, and add the connector to your AI app. Price: MCP-only is CHF 9 a month or CHF 69 a year, after a 7-day free trial ( plans ). Coaching is CHF 29.90 a month. Best for: athletes who want a coach that gets in touch first, as well as answers. 3. Tredict Tredict is an endurance training platform with its own MCP server. It imports up to 8 years of Concept2 sessions, but hand-entered Logbook results do not come across. It works best if Tredict is already your training log, and its MCP server needs Tredict's paid access for detailed data. Setup: create a Tredict account, connect Concept2, then add Tredict to your AI app. Price: 2 months free, then $49 for 12 months, paid once (€49 in Europe; prices ). Best for: endurance athletes who already plan their training in Tredict. 4. IcuSync, through Intervals.icu An indirect route. The Logbook syncs to Intervals.icu , which is free: the last 28 days first, then your full history on request. IcuSync serves Intervals.icu over MCP in Claude on the web, desktop, and phone, and in ChatGPT and Grok. Setup: create an Intervals.icu account, link Concept2 there, sign up for IcuSync, and paste its URL into your AI app. Price: Intervals.icu is free; IcuSync is $20 a year. Best for: athletes who already train in Intervals.icu. 5. concept2-mcp A small local server with 5 read tools , stroke data among them, MIT licensed. It runs with one npx command and your personal token. It is a one-day project with no stars yet, so expect little maintenance. Setup: moderate. Create your personal token in the Logbook, install Node, and add its npx command with the token to your AI app's config. Desktop apps only. Price: free and open source. Best for: rowers who want stroke-level detail on their own computer. 6. mcp-concept2-logbook A local server with 10 tools , read-only by default, MIT licensed. It takes your personal token or your own Concept2 sign-in app, which renews itself. Setup: moderate. Create a token or register an app in the Logbook, install Node, and add its npx command to your AI app's config. Desktop apps only. Price: free and open source. Best for: people who want a broader local tool set. 7. ergmcp A Python server with 6 tools , personal bests among them, MIT licensed. It runs with uv and your personal token. Setup: moderate. Create your personal token, install uv, and add its command with the token to your AI app's config. Desktop apps only. Price: free and open source. Best for: Python users who want a small server with personal bests. 8. Intervals.icu with a local MCP server The do-it-yourself version of the Intervals.icu route: the Logbook syncs to Intervals.icu, and hhopke/intervals-icu-mcp reads it from your own computer with 67 tools, MIT licensed. It sees what Intervals.icu imported. Setup: moderate. Link Concept2 to Intervals.icu, install uv, copy your athlete ID and API key, and add one command to your AI app's config. Desktop apps only. Price: free and open source. Best for: people who already train in Intervals.icu and want a local server. We left out two more local servers, because neither has a license file and one includes a tool that deletes a Logbook result for good. Concept2 MCP options at a glance (prices as of September 2026) Tool Setup Reads from Price freddy None; web, desktop, and phone The Logbook, directly Free tier; $49/year AthleteData Account; web, desktop, and phone The Logbook, directly CHF 69/yr or CHF 9/mo Tredict Tredict account; web, desktop, and phone Tredict's import 2 months free, then $49 per 12 months IcuSync Two accounts; web, desktop, and phone Intervals.icu's import $20/yr concept2-mcp Token and config file; desktop only The Logbook, directly Free mcp-concept2-logbook Token or app, and config file; desktop only The Logbook, directly Free ergmcp Token and config file; desktop only The Logbook, directly Free hhopke/intervals-icu-mcp Terminal and config file; desktop only Intervals.icu's import Free How to choose For most people, the answer is freddy: every Logbook result, nothing to run, and your erg sessions next to your recovery data in any AI app that speaks MCP. The exceptions: You want a coach that gets in touch first: AthleteData. You already train in Intervals.icu: IcuSync. You want stroke-level detail on your own computer: concept2-mcp. Prices and details verified in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Health Connect MCP servers: 5 ways to get your Health Connect data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/health-connect-mcp-servers
Five ways to connect Android's Health Connect to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with how fresh the data is and a price for each.
Health Connect collects your sleep, steps, workouts, and heart rate from the apps and watches on your Android phone, and the AI assistant you talk to every day cannot see any of it. The Model Context Protocol (MCP) is the open standard that connects the two. For Health Connect the choice is small: five options work today. This list ranks them by setup effort, from least to most. freddy is ours, and it is first: one Android app, nothing to keep open, and it works in any AI app that speaks MCP. Each entry says who it suits. What to know first Health Connect has no cloud API. It lives on your phone, and data leaves it only through an Android app you give permission to, or through Health Connect's own export file. So the question that sorts the options is freshness: does an app upload your data in the background, read it live while you ask, or wait for a daily export? Two Android permissions decide how much an app can see. Without the history permission, an app reads only the 30 days before you first grant access; without the background permission, it reads only while it is open ( Android docs ). Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. Claude's Android app can also read Health Connect without MCP, in beta for Pro and Max plans in the US, on Android 14 or later ( Claude help ). It works inside that one app only. Outside the US, or in more than one AI app, you need one of the options below. The list 1. freddy freddy is a hosted MCP server that brings your health data from all your apps into one place. You install freddy for Android , allow it to read Health Connect, and add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. freddy can send new data without being open and reach back further than 30 days. Your phone sends new data in the background when Android allows it, and right away when you open freddy. Because freddy is hosted, you can ask from your laptop or your phone. Health Connect already carries data from apps like Samsung Health, Google Health, and many others, and freddy also connects Garmin, Oura, Polar, and more directly. Our setup guide walks through every step. Setup: install the Android app, allow Health Connect access, and add freddy to your AI app. No computer, no export files. Price: free for one source with a short history window, or $49/year for all your sources and your full history. Best for: almost everyone with an Android phone. 2. Wellness Project Wellness Project is a coaching app for Android with a hosted MCP server. The app reads Health Connect, and its server lists 73 tools, some of which write to Wellness Project. Its pages do not say whether it syncs in the background or asks for the history permission. The free plan allows 3 analysis chats a day, with the app and your AI app counted together. Setup: install the Android app, allow Health Connect access, then add its MCP address to your AI app and sign in. Price: a free Basic plan. Pro is $9.99 a month or $99.99 a year, with an introductory $4.99 a month or $39.99 a year until October 15 ( pricing ). Best for: people who want a coaching app alongside their AI assistant. 3. Genetrix16/health-connect-mcp A small open-source pair: an Android app and a server on your computer, MIT licensed, with 7 read-only tools. Your computer asks the phone over your Wi-Fi, so the data is live on every question. The app does not ask for the history permission, so it sees only the 30 days before you first grant access. Its README install command points at a package that does not exist on npm, so you run the server from the repository. Setup: involved. Build or install the Android app, clone the repository, run the server, and add it to your desktop AI app. Phone and computer must share a Wi-Fi network. Price: free and open source. Best for: tinkerers who want live data at their desk. 4. Open Wearables An open-source platform that you run on your own server, with 2,562 stars and an MIT license. Since version 0.9.0 it has a Health Connect provider, and its Android app asks for the history and background permissions. It includes an MCP server with 6 tools and covers several other wearables. The Android app is a beta; you ask for it on the project's Discord. Setup: the most work of the ready-made options. Run its Docker stack, get the beta Android app, connect it, create an API key, and set up the MCP server. Desktop apps only. Price: free software. You pay for the server. Best for: developers who want several sources on hardware they control. 5. pricejoshua/health-connect-mcp A homelab project built on Health Connect's export file : Health Connect saves a daily export to Google Drive, a sync tool copies it to your server, and an MCP server with 11 tools, raw SQL among them, reads it behind a Cloudflare tunnel. Because it runs on the web, it works in the Claude phone app. Your data is a day old at best, and the repository has no license file, so its terms are unclear. Setup: the most work on this list. Turn on Health Connect's scheduled export, set up the file sync, the database, the server, and the tunnel. Price: free. Best for: homelab owners who want their own copy of the export. Health Connect MCP options at a glance (prices as of September 2026) Tool Data freshness Setup Price freddy Background sync, and on app open Android app; web, desktop, and phone Free tier; $49/year Wellness Project Sync, interval not stated Android app; web, desktop, and phone Free (3 chats a day); Pro $9.99/mo Genetrix16 Live, over Wi-Fi Android app and local server; desktop only Free Open Wearables Background sync (beta app) Your own server; desktop only Free; you pay for hosting pricejoshua Daily export Export, sync, database, and tunnel Free How to choose For most people, the answer is freddy: one Android app, history included, nothing to keep open, and it works in any AI app that speaks MCP, on your phone too. The exceptions: You want a coaching app alongside: Wellness Project. You want live data at your desk and like to tinker: Genetrix16/health-connect-mcp. You want several sources on hardware you control: Open Wearables. Prices and details verified in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Hevy MCP servers: 8 ways to get your Hevy data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/hevy-mcp-servers
Eight ways to connect Hevy to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with prices and which ones need Hevy Pro.
Hevy holds every set, rep, and weight you have logged, and the AI assistant you talk to every day cannot see any of it. The Model Context Protocol (MCP) is the open standard that connects the two, and eight options now do it for Hevy. This list ranks them by setup effort, from least to most. freddy is ours, and it is first: nothing to install, and your lifting next to your sleep and recovery apps in any AI app that speaks MCP. Each entry says who it suits. What to know first The question that sorts Hevy options is Hevy Pro. Hevy's public API works with a personal API key, and the key is only for Pro members ( Hevy API ). Pro costs $2.99 a month, $23.99 a year, or $74.99 once ( pricing ), and the key is in Hevy's settings under Developer, with no approval step. Without Pro, the one route is Hevy's own ChatGPT plugin, which is free and mostly builds routines. Hevy's API can create and update workouts and routines, but it has no delete and no undo, and an update replaces the whole record. If you let an AI write to Hevy, check what it changes. Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. The list 1. freddy freddy is a hosted MCP server that brings your health data from all your apps into one place. You sign in with a code we email you, enter your Hevy API key one time, and add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. freddy reads your workouts and body measurements, history included, and never writes to Hevy, so your log stays as you left it. The same account reads Garmin, Oura, Polar, Apple Health, and more, so a question like "do my lifts drop in weeks I sleep badly" works across all of them. Our Hevy guide walks through every step. Setup: three steps in your browser, plus your Hevy API key, which needs Hevy Pro. Nothing to install, and it works on your phone too. Price: free for one source with a short history window, or $49/year for all your sources and your full history. Best for: Hevy Pro members who want their training next to their sleep and recovery, without running anything. 2. Hevy for ChatGPT Hevy's own ChatGPT plugin signs in with your Hevy account, and Hevy says it works with free Hevy and free ChatGPT accounts. It reads your workouts and creates routines, folders, and plans; it does not log or edit workouts, and it does not read body measurements. It works in ChatGPT only. Hevy's free plan allows 4 routines, which a generated plan can fill quickly. Setup: install Hevy from ChatGPT's Plugin Directory and sign in with Hevy. Price: free. Best for: ChatGPT users without Hevy Pro who want the AI to build routines. 3. hevy-mcp, hosted The best-known community project also runs as a free hosted server . It reads and writes workouts, routines, folders, body measurements, and custom exercises, and you enter your Pro key on its sign-in page. The code is open, with 483 stars and an MIT license. Setup: add its URL as a custom connector in Claude and enter your Pro key on its page. In ChatGPT it needs developer mode. Price: free and open source. Best for: Hevy Pro members who want the AI to write routines and workouts at no cost. 4. Wellness Project Wellness Project is a hosted coaching app with its own MCP server. It syncs Hevy every hour with your full history, never writes to Hevy, and labels its Hevy connection as alpha. The free plan allows 3 analysis chats a day, with the app and your AI app counted together. Setup: paste your Hevy API key in the app, then add its MCP address to your AI app and sign in. Price: a free Basic plan. Pro is $9.99 a month or $99.99 a year, with an introductory $4.99 a month or $39.99 a year until October 15 ( pricing ). Best for: people who want a coaching app alongside their AI assistant. 5. AthleteData AthleteData is a hosted AI coach with a cheaper MCP-only plan. It lists 18 Hevy tools that read and write, next to more than 30 other apps. The coaching plan messages you after workouts; the MCP-only plan leaves that out. Setup: start the trial, paste your Hevy API key in the dashboard, and add the connector to your AI app. Price: MCP-only is CHF 9 a month or CHF 69 a year, after a 7-day free trial ( plans ). Coaching is CHF 29.90 a month. Best for: athletes who want a coach that gets in touch first, as well as answers. 6. FitnessPeak FitnessPeak , from a Norwegian company, runs a hosted MCP server that covers Hevy, WHOOP, and Withings, with 7 read-only Hevy tools. It connects with a key, so it suits Claude, Cursor, and VS Code; its own docs say ChatGPT's connector screen does not work with it. Setup: sign up with a card, connect Hevy, and add its key to your AI app. Price: $39 a year, after a 14-day trial ( pricing ). Best for: people who want Hevy, WHOOP, and Withings in one small hosted server. 7. hevy-mcp, local The same chrisdoc/hevy-mcp project, run on your own computer with your API key in its config. It is the most active Hevy server, with 24 contributors and releases this month. Setup: moderate. Install Node 20 and add one command with your Hevy API key to your AI app's config. Desktop apps only. Price: free and open source. Best for: people who would rather keep their key on their own computer. 8. khglynn/hevy-mcp A server you deploy yourself on Cloudflare Workers: 25 tools , 8 of which write, and each connection can be read-only or read-write. It is new, credits chrisdoc's project, and works in Claude on the web and phone because it runs on the web. Setup: involved. You need a Cloudflare account, Node 22, a storage namespace, secrets, and a deploy. Then you add your own URL to your AI app. Price: free, within Cloudflare's free tier. Best for: developers who want a server they own that also works on their phone. Hevy MCP options at a glance (prices as of September 2026) Tool Setup Needs Hevy Pro Price freddy API key; web, desktop, and phone Yes Free tier; $49/year Hevy for ChatGPT None; ChatGPT only No Free hevy-mcp, hosted API key; web, desktop, and phone Yes Free Wellness Project App and connector; web, desktop, and phone Yes Free (3 chats a day); Pro $9.99/mo AthleteData API key; web, desktop, and phone Yes CHF 69/yr or CHF 9/mo FitnessPeak Key; desktop apps and claude.ai Yes $39/yr hevy-mcp, local Terminal and config file; desktop only Yes Free khglynn/hevy-mcp Cloudflare deploy; web, desktop, and phone Yes Free (Cloudflare free tier) How to choose For most Hevy Pro members, the answer is freddy: your lifting next to your recovery data, nothing to run, and it never changes your log. The exceptions: You do not pay for Hevy Pro and use ChatGPT: Hevy's own ChatGPT plugin. You want the AI to write routines and workouts: hevy-mcp, hosted or local. You want a coach that gets in touch first: AthleteData. You want a server you own that works on your phone: khglynn/hevy-mcp. Prices and details verified in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Intervals.icu MCP servers: 8 ways to get your Intervals.icu data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/intervals-icu-mcp-servers
Eight ways to connect Intervals.icu to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with prices and which ones can plan workouts.
Intervals.icu already holds your training load, fitness curve, wellness notes, and every workout your watch or bike computer sends it. The Model Context Protocol (MCP) is the open standard that lets the AI assistant you talk to every day read all of that, and eight options now do it for Intervals.icu. This list ranks them by setup effort, from least to most. freddy is ours, and it is first: nothing to install, no API key in a file, and Intervals.icu next to your sleep and recovery apps in any AI app that speaks MCP. Each entry says who it suits. What to know first Intervals.icu gives every athlete a personal API key, free, under Developer Settings on the settings page. The key opens your whole account, so it belongs in a config file on your own computer, never in a chat or a hosted form. Hosted services use Intervals.icu's own sign-in screen instead, which you can revoke per app on the same settings page ( API docs ). One limit applies to every option on this list: activities that reach Intervals.icu through its Strava sync do not come back out through the API. Intervals.icu confirmed this in October 2025 ( forum ). If Strava is your only link, connect your watch to Intervals.icu directly, or use its import of your Strava archive, which is a Supporter feature. Intervals.icu itself is free, with an optional Supporter plan at $4 a month ( pricing ). It has no official MCP server yet; in September 2026 its developer said one is at the spec stage ( forum ). Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. The list 1. freddy freddy is a hosted MCP server that brings your health data from all your apps into one place. You sign in with a code we email you, connect Intervals.icu on its own sign-in screen, and add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. New activities and wellness entries arrive by push soon after Intervals.icu gets them, and your history comes with them. The same account reads Garmin, Oura, Polar, Withings, Apple Health, and more, so a question like "does my HRV drop the day after a hard interval session" works across your training and your recovery. Our Intervals.icu guide walks through every step. freddy reads your data; it does not write workouts to your Intervals.icu calendar. If you want the AI to plan sessions for you, look at entries 2, 4, 5, and 7. Setup: three steps in your browser. Nothing to install, no API key to copy, and it works on your phone too. Price: free for one source with a short history window, or $49/year for all your sources and your full history. Best for: athletes who want their training next to their sleep and recovery data, without running anything. 2. icuvisor icuvisor is an open-source server with a free hosted version. It signs in with Intervals.icu's own screen and offers 73 tools in three sizes, 51 of which read and the rest write or delete. The hosted version is read-only by default, and you turn on writes yourself. It works in Claude, in ChatGPT through developer mode, and in Claude Desktop. The hosted version's price is not written down anywhere except a forum post by its maintainer that calls it free. Setup: add its URL as a connector, pick a tool set, and approve access on Intervals.icu. A local version also installs with one command and your API key. Price: free and open source (MIT). Best for: athletes who want the AI to read and plan in Intervals.icu at no cost. 3. gregaria-mcp gregaria-mcp is a small, free hosted relay built for Claude. It signs in with Intervals.icu's own screen and needs no account of its own. It reads your data and can write wellness entries, calendar events, and activity messages. It is a one-developer project with a handful of GitHub stars, and ChatGPT support is not there yet. Setup: add its URL as a custom connector in Claude and log in to Intervals.icu. Price: free and open source (MIT). Best for: Claude users who want a free hosted connection with nothing to sign up for. 4. IcuSync IcuSync is a paid hosted server from an Australian company, with 24 tools that read and write: workouts, wellness, FTP, and activity comments, plus sharing with a coach. It signs in with Intervals.icu's own screen and gives you a personal URL. It works in Claude on every plan, on the web, desktop, and phone, in ChatGPT through developer mode, and in Grok. Setup: sign up, connect Intervals.icu, and paste your personal URL into your AI app. Price: $20 a year. The launch free trial has closed. Best for: athletes who want a maintained, paid service that plans workouts with them. 5. easy-intervals easy-intervals is a hosted server from Germany with 33 tools that read and write events, wellness, and activities. It signs in with Intervals.icu's own screen and works in Claude, ChatGPT, Gemini, Grok, and Cursor. The free plan gives 7 days of full access, then 15 requests a day. Setup: register, connect Intervals.icu, add its URL as a connector, and sign in. Price: free with a daily limit, or €18 a year for Power, which does not renew by itself. Best for: people who want a low-cost yearly plan and several AI apps. 6. AthleteData AthleteData is a hosted AI coach with a cheaper MCP-only plan. For Intervals.icu it offers 11 tools, 3 of which write to your calendar, next to Garmin, WHOOP, Oura, and others. The coaching plan messages you after workouts; the MCP-only plan leaves that out. Setup: start the trial, link Intervals.icu, then add the connector to your AI app. Price: MCP-only is CHF 9 a month or CHF 69 a year, after a 7-day free trial ( plans ). Coaching is CHF 29.90 a month. Best for: athletes who want a coach that gets in touch first, as well as answers. 7. hhopke/intervals-icu-mcp The most complete local server: 67 tools , plus resources and prompts, MIT licensed. It reads and writes, and deletes are off unless you allow them. It is the maintained continuation of an earlier project by eddmann, and says so. Setup: moderate. Install uv, copy your athlete ID and API key from Intervals.icu, and add one command with those two values to your AI app's config. Desktop apps only. Price: free and open source. Best for: people at home in a terminal who want every tool on their own computer. 8. mvilanova/intervals-mcp-server The most-starred Intervals.icu MCP server, with 361 stars , about 20 tools, and a GPL-3.0 license. It reads your data and writes events, notes, and activity messages. Its ChatGPT recipe runs the server behind a public tunnel with no sign-in, so anyone with the address could read your data. Keep it to Claude Desktop. Setup: involved. Install uv, clone the repository, create its environment, put your athlete ID and API key in a settings file, and register it with Claude Desktop. Desktop apps only. Price: free and open source. Best for: developers who want the best-known project and will run it locally. Intervals.icu MCP options at a glance (prices as of September 2026) Tool Setup Intervals.icu sign-in Price freddy None; web, desktop, and phone Official sign-in Free tier; $49/year icuvisor None (hosted) or one command (local) Official sign-in, or your API key locally Free gregaria-mcp None; Claude Official sign-in Free IcuSync Account; web, desktop, and phone Official sign-in $20/yr easy-intervals Account; web, desktop, and phone Official sign-in Free with a daily limit; €18/yr AthleteData Account; web, desktop, and phone Official sign-in CHF 69/yr or CHF 9/mo hhopke/intervals-icu-mcp Terminal and config file; desktop only Your API key Free mvilanova/intervals-mcp-server Clone and settings file; desktop only Your API key Free How to choose For most people, the answer is freddy: no API key in a file, nothing to keep running, and Intervals.icu next to your recovery data in any AI app that speaks MCP. The exceptions: You want the AI to plan workouts in Intervals.icu at no cost: icuvisor. You want a paid service that plans with you: IcuSync, or easy-intervals for the lowest yearly price. You want a coach that gets in touch first: AthleteData. You want everything on your own computer: hhopke/intervals-icu-mcp. Whatever you pick, never put your Intervals.icu API key into a chat or a hosted form. It opens your whole account until you make a new one. Prices and details verified in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Polar MCP servers: 8 ways to get your Polar data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/polar-mcp-servers
Eight ways to connect your Polar watch to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with prices and how much history each reaches.
Your Polar watch records your training, sleep, and Nightly Recharge every day, and the AI assistant you talk to every day cannot see any of it. The Model Context Protocol (MCP) is the open standard that connects the two, and eight options now do it for Polar. This list ranks them by setup effort, from least to most. freddy is ours, and it is first: nothing to install, no developer client to register, and it reaches back into your Polar history. Each entry says who it suits. What to know first The question that sorts Polar options is history. Polar's standard API, AccessLink v3, returns only workouts that sync after you connect a tool, and only for 30 days; sleep and Nightly Recharge reach back 28 days ( AccessLink v3 ). Polar's newer API, v4, can ask for older data in date ranges ( AccessLink v4 ), and Polar's full data export covers everything as JSON files. Any Polar Flow user can register a free developer client at Polar's developer portal , so every local option below starts there. Hosted services use their own approved client, so you only approve access on Polar's screen. One naming trap: the "Polar" connector in the Claude and ChatGPT directories is Polar Analytics, an ecommerce tool, not your Polar watch. Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. The list 1. freddy freddy is a hosted MCP server that brings your health data from all your apps into one place. You sign in with a code we email you, connect Polar on Polar's own sign-in screen, and add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. New workouts, sleep, and Nightly Recharge arrive by push soon after your watch syncs. A second Polar approval in freddy brings in your older history. The same account reads Garmin, Oura, Withings, Apple Health, and more, so you can ask across all of them. Our Polar guide walks through every step. Setup: three steps in your browser, plus one more approval on Polar's screen for your older history. Nothing to install, and it works on your phone too. Price: free for one source with a short history window, or $49/year for all your sources and your full history. Best for: almost everyone with a Polar watch, and especially people who want their history, not just what comes next. 2. Tredict Tredict is an endurance training platform with its own MCP server and ChatGPT plugin. It connects Polar Flow for workouts, nightly HRV, and sleep, and it can turn a Polar data export into history. From the chat, it can build a training plan in Tredict. It works best if Tredict is already your training log, and its MCP server needs Tredict's paid access for detailed data. Setup: create a Tredict account, connect Polar Flow, then add Tredict to your AI app. Price: 2 months free, then $49 for 12 months, paid once (€49 in Europe; prices ). Best for: endurance athletes who already plan their training in Tredict. 3. Wellness Project Wellness Project is a hosted coaching app with its own MCP server. It connects Polar directly for activity, sleep, recovery, heart, and training data, and its server lists 73 tools, some of which write to Wellness Project. The free plan allows 3 analysis chats a day, with the app and your AI app counted together. Its pages do not say how much Polar history it brings in. Setup: sign in with Apple or Google, connect Polar in the app, then add its MCP address to your AI app and sign in. Price: a free Basic plan. Pro is $9.99 a month or $99.99 a year, with an introductory $4.99 a month or $39.99 a year until October 15 ( pricing ). Best for: people who want a coaching app alongside their AI assistant. 4. AthleteData AthleteData is a hosted AI coach with a cheaper MCP-only plan. It connects Polar for activities, sleep, daily activity, and Nightly Recharge, next to more than 30 other apps. The coaching plan messages you after workouts; the MCP-only plan leaves that out. Setup: start the trial, connect Polar, copy the API key from your dashboard, and add it to your AI app. Price: MCP-only is CHF 9 a month or CHF 69 a year, after a 7-day free trial ( plans ). Coaching is CHF 29.90 a month. Best for: athletes who want a coach that gets in touch first, as well as answers. 5. IcuSync, through Intervals.icu An indirect route. Polar syncs to Intervals.icu , which is free, with activities, sleep, HRV, and resting heart rate. IcuSync then serves Intervals.icu over MCP, with tools that read and plan, in Claude on any plan and in ChatGPT and Grok. Connect Polar to Intervals.icu directly, not through Strava: activities that arrive through Strava do not come back out of the Intervals.icu API. Setup: moderate. Create an Intervals.icu account, connect Polar there, sign up for IcuSync, and paste its URL into your AI app. Price: Intervals.icu is free; IcuSync is $20 a year. Best for: athletes who already train in Intervals.icu. 6. polar-mcp-unofficial A local server on Polar's newer API: 38 tools , MIT licensed, published on npm as polar-mcp-unofficial. It reads your data, and it hides GPS positions by default. Setup: involved. Register your own Polar developer client with a local redirect address, install Node 20, and run its setup, auth, and doctor commands, which write your Claude Desktop config for you. Desktop apps only. Price: free and open source. Best for: technical Polar users who want the newer API on their own computer. 7. chanezgr/polar_mcp A Python server on Polar's standard API with 15 tools , a local SQLite cache, and saved FIT files, under GPL-3.0. A scheduled mode keeps collecting, so your history grows from the day you start it. It has no sign-in helper: you get a Polar access token yourself. Setup: the most of the local options. Register your own Polar client, get an access token by hand, and run it in Docker or with pip. Price: free and open source. Best for: developers who want a growing local archive of their Polar data. 8. Open Wearables An open-source platform that you run on your own server, with 2,562 stars and an MIT license. It connects Polar with your own developer client, receives Polar's update notifications, and includes an MCP server with 6 tools. It covers several other wearables too. Setup: the most work on this list. Run its Docker stack, put your Polar client's keys in its config, and set up the MCP server. Desktop apps only. Price: free software. You pay for the server. Best for: developers who want several sources on hardware they control. Polar MCP options at a glance (prices as of September 2026) Tool Setup Polar history Price freddy None; web, desktop, and phone Standard API plus older history through Polar's newer API Free tier; $49/year Tredict Tredict account; web, desktop, and phone Synced data, plus a Polar export upload 2 months free, then $49 per 12 months Wellness Project App and connector; web, desktop, and phone Not stated Free (3 chats a day); Pro $9.99/mo AthleteData API key; web, desktop, and phone Not stated CHF 69/yr or CHF 9/mo IcuSync via Intervals.icu Two accounts; web, desktop, and phone What Intervals.icu has synced $20/yr polar-mcp-unofficial Your own Polar client; desktop only Newer API, by date range Free chanezgr/polar_mcp Your own client and token; desktop only Grows from the day you start Free Open Wearables Your own server and Polar client; desktop only Standard API Free; you pay for hosting How to choose For most people, the answer is freddy: no developer client, nothing to keep running, your older Polar history, and it works in any AI app that speaks MCP, on your phone too. The exceptions: You already plan your training in Tredict: its own connector. You want a coach that gets in touch first: AthleteData, or Wellness Project for a coaching app. You already train in Intervals.icu: IcuSync. You want it on your own computer: polar-mcp-unofficial. Prices and details verified in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Suunto MCP servers: 9 ways to get your Suunto data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/suunto-mcp-servers
Nine ways to connect your Suunto watch to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with prices and who each one suits.
Your Suunto watch records your runs, rides, dives, and daily recovery, and the AI assistant you talk to every day cannot see any of it. The Model Context Protocol (MCP) is the open standard that connects the two, and nine options now do it for Suunto. This list ranks them by setup effort, from least to most. freddy is ours, and it is first: nothing to install, and Suunto next to your other apps in any AI app that speaks MCP. Each entry says who it suits. What to know first The question that sorts Suunto options is simple: does the tool connect to Suunto for you, or do you set up the connection yourself? Suunto also syncs your workouts to partner platforms, which opens indirect routes: Suunto to Intervals.icu, then an Intervals.icu MCP server. Those routes carry what the partner imports. One tool on this list signs in with your Suunto email and password through the private API behind the Suunto app. It can stop working at any time, so use it with care. Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. The list 1. freddy freddy is a hosted MCP server that brings your health data from all your apps into one place. You sign in with a code we email you, connect Suunto on Suunto's own sign-in screen, and add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. freddy connects to Suunto directly, so there is nothing to apply for. New workouts arrive by push soon after your watch syncs, and your history comes with them. The same account reads Garmin, Oura, Polar, Apple Health, and more, so you can ask across all of them. Our Suunto guide walks through every step. Setup: three steps in your browser. Nothing to install, nothing to update, and it works on your phone too. Price: free for one source with a short history window, or $49/year for all your sources and your full history. Best for: almost everyone with a Suunto watch, and especially people who use more than one app. 2. Tredict Tredict is an endurance training platform with its own MCP server and ChatGPT plugin. It imports Suunto workouts, nightly HRV, and sleep, up to 2 years back, and from the chat it can build a training plan in Tredict. It works best if Tredict is already your training log, and its MCP server needs Tredict's paid access for detailed data. Setup: create a Tredict account, connect Suunto, then add Tredict to your AI app. Price: 2 months free, then $49 for 12 months, paid once (€49 in Europe; prices ). Best for: endurance athletes who already plan their training in Tredict. 3. AthleteData AthleteData is a hosted AI coach with a cheaper MCP-only plan. It reads Suunto activities, sleep, and daily activity, next to more than 30 other apps. The coaching plan messages you after workouts; the MCP-only plan leaves that out. Setup: sign up, connect Suunto, then add its connector to your AI app and sign in. Price: MCP-only is CHF 9 a month or CHF 69 a year, after a 7-day free trial ( plans ). Coaching is CHF 29.90 a month. Best for: athletes who want a coach that gets in touch first, as well as answers. 4. STAS, through Intervals.icu STAS is a free AI coaching layer that reads Intervals.icu, where Suunto syncs your workouts and, since 2025, sleep, HRV, and resting heart rate. It can change your plan in Intervals.icu after you approve the change, and it works as a Claude connector or as a GPT in ChatGPT. Setup: create a free Intervals.icu account, link Suunto there, sign in to STAS with Intervals.icu, and add its connector. Price: free. Best for: athletes who want a free coach on top of Intervals.icu. 5. IcuSync, through Intervals.icu IcuSync is a paid hosted server for Intervals.icu, with tools that read your data and send planned workouts to your watch. With Suunto linked to Intervals.icu, it reaches your Suunto sessions in Claude on the web, desktop, and phone, and in ChatGPT and Grok. Setup: create an Intervals.icu account, link Suunto there, sign up for IcuSync, and paste its URL into your AI app. Price: Intervals.icu is free; IcuSync is $20 a year. Best for: athletes who already train in Intervals.icu. 6. Intervals.icu with a local MCP server The do-it-yourself version of the same route: Suunto syncs to Intervals.icu, and hhopke/intervals-icu-mcp reads it from your own computer with 67 tools, MIT licensed. Intervals.icu's personal API key is free. Setup: moderate. Link Suunto to Intervals.icu, install uv, copy your athlete ID and API key, and add one command to your AI app's config. Desktop apps only. Price: free and open source. Best for: people at home in a terminal who want every tool on their own computer. 7. suuntool A Go command-line tool with a built-in MCP server, read-only by default , MIT licensed, for macOS and Linux. It signs in with your Suunto email and password through the private API behind the Suunto app. It stops working if Suunto changes that private API. Setup: moderate. Install it, sign in with your Suunto account, and add its MCP command to your AI app. Desktop apps only. Price: free and open source. Best for: technical users who accept that risk for full access. 8. googlarz/suunto-mcp A TypeScript server on Suunto's official API: about 23 tools , MIT licensed and active, including FIT uploads and SuuntoPlus Guides. Setup: involved. Get Suunto API access, register an app, clone and build the server, fill in its settings, sign in, and add it to Claude Desktop. Price: free and open source. Best for: developers who have their own Suunto API access. 9. Open Wearables An open-source platform that you run on your own server, with 2,562 stars and an MIT license. It includes an MCP server with 6 tools and covers several other wearables. For Suunto, you use your own Suunto API access. Setup: the most work on this list. Get Suunto API access, run its Docker stack, add your Suunto keys, and set up the MCP server. Desktop apps only. Price: free software. You pay for the server. Best for: developers with Suunto API access who want several sources on hardware they control. Suunto MCP options at a glance (prices as of September 2026) Tool Setup Suunto access Price freddy None; web, desktop, and phone Official Free tier; $49/year Tredict Tredict account; web, desktop, and phone Official 2 months free, then $49 per 12 months AthleteData Account; web, desktop, and phone Official sign-in CHF 69/yr or CHF 9/mo STAS Two accounts; Claude and ChatGPT Through Intervals.icu Free IcuSync Two accounts; web, desktop, and phone Through Intervals.icu $20/yr hhopke/intervals-icu-mcp Terminal and config file; desktop only Through Intervals.icu Free suuntool Terminal; desktop only Password, private API Free googlarz/suunto-mcp Developer setup; desktop only Official, your own app Free Open Wearables Your own server; desktop only Official, your own app Free; you pay for hosting How to choose For most people, the answer is freddy: nothing to keep running, and Suunto next to your other apps in any AI app that speaks MCP, on your phone too. The exceptions: You already plan your training in Tredict: its own connector. You already train in Intervals.icu: STAS for free, or IcuSync. You want a coach that gets in touch first: AthleteData. You have your own Suunto API access: googlarz/suunto-mcp. Prices and details verified in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Ultrahuman MCP servers: 5 ways to get your Ultrahuman Ring data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/ultrahuman-mcp-servers
Five ways to connect your Ultrahuman Ring and M1 to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with prices and glucose support.
Your Ultrahuman Ring tracks your sleep, recovery, and movement every day, and an M1 sensor adds your glucose, but the AI assistant you talk to every day cannot see any of it. The Model Context Protocol (MCP) is the open standard that connects the two, and five options work for Ultrahuman today. This list ranks them by setup effort, from least to most. freddy is ours, and it is first: nothing to install, no token to paste, M1 glucose included, and it works in any AI app that speaks MCP. Each entry says who it suits. What to know first Ultrahuman offers two official ways in, both on its developer page : an OAuth sign-in for apps, and a personal API token for your own data. Hosted services use the sign-in, so you only approve access on Ultrahuman's screen. A server you run yourself uses a token. Never share your token or paste it into someone else's website. The second question is glucose. If you wear an M1 sensor, check that the tool reads it: not every one does. Ultrahuman's own AI, Jade, works inside the Ultrahuman app only. Ultrahuman ships no MCP server of its own. Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. The list 1. freddy freddy is a hosted MCP server that brings your health data from all your apps into one place. You sign in with a code we email you, connect Ultrahuman on Ultrahuman's own sign-in screen, and add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. freddy reads your ring's sleep, recovery, heart rate, HRV, temperature, and movement, plus M1 glucose, with your history. A night's numbers fill in after you wake and your ring syncs. The same account reads Garmin, Oura, Polar, Apple Health, and more. Our Ultrahuman guide walks through every step. Setup: three steps in your browser. Nothing to install, nothing to update, and it works on your phone too. Price: free for one source with a short history window, or $49/year for all your sources and your full history. Best for: almost everyone with an Ultrahuman Ring, and especially M1 wearers. 2. Wellness Project Wellness Project is a hosted coaching app with its own MCP server. It signs in to Ultrahuman directly, reads ring data and M1 glucose, and its server lists 73 tools, some of which write to Wellness Project, never to Ultrahuman. The free plan allows 3 analysis chats a day, with the app and your AI app counted together. Setup: create an account, authorize Ultrahuman in its settings, then add its MCP address to your AI app and sign in. Price: a free Basic plan. Pro is $9.99 a month or $99.99 a year, with an introductory $4.99 a month or $39.99 a year until October 15 ( pricing ). Best for: people who want a coaching app alongside their AI assistant. 3. AthleteData AthleteData is a hosted AI coach with a cheaper MCP-only plan. It signs in to Ultrahuman for your ring data, next to more than 30 other apps, but it does not read M1 glucose. The coaching plan messages you after workouts; the MCP-only plan leaves that out. Setup: start the trial, connect Ultrahuman, and add the connector to your AI app. Price: MCP-only is CHF 9 a month or CHF 69 a year, after a 7-day free trial ( plans ). Coaching is CHF 29.90 a month. Best for: athletes who want a coach that gets in touch first, and do not need glucose. 4. Fulcra Context, through Apple Health Fulcra Context is a personal data store with a hosted MCP server. Its iPhone app reads Apple Health, and the Ultrahuman app writes HRV, temperature, energy, and M1 glucose there. Fulcra's pages do not name Ultrahuman; the route works because Fulcra reads what Apple Health holds. It is iPhone only, and it sees only what Ultrahuman shares with Apple Health. Setup: moderate. Turn on Apple Health sync in the Ultrahuman app, install Context, allow Apple Health access, then add Fulcra's server to your AI app. Price: Personal is free. Pro is $14.99 a month or $149.99 a year ( pricing ). Best for: iPhone users who want a broader personal data store. 5. Open Wearables An open-source platform that you run on your own server, with 2,562 stars and an MIT license. It signs in to Ultrahuman with an OAuth app you create on Ultrahuman's developer page, reads ring data and M1 glucose every hour, and includes an MCP server with 6 tools. Setup: the most work on this list. Run its Docker stack, create your own Ultrahuman OAuth app, put its keys in the config, connect in its dashboard, and set up the MCP server. Desktop apps only. Price: free software. You pay for the server. Best for: developers who want several sources on hardware they control. We left out the small open-source Ultrahuman servers. Ultrahuman MCP options at a glance (prices as of September 2026) Tool Setup M1 glucose Price freddy None; web, desktop, and phone Yes Free tier; $49/year Wellness Project App and connector; web, desktop, and phone Yes Free (3 chats a day); Pro $9.99/mo AthleteData Account; web, desktop, and phone No CHF 69/yr or CHF 9/mo Fulcra Context iPhone app; web, desktop, and phone Through Apple Health Free; Pro $14.99/mo Open Wearables Your own server and OAuth app; desktop only Yes Free; you pay for hosting How to choose For most people, the answer is freddy: no token, nothing to keep running, M1 glucose included, and it works in any AI app that speaks MCP, on your phone too. The exceptions: You want a coaching app alongside: Wellness Project. You want a coach that gets in touch first, and no glucose: AthleteData. You want several sources on hardware you control: Open Wearables. Whatever you pick, keep your Ultrahuman token off other people's websites. Ultrahuman's terms forbid sharing it. Prices and details verified in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Wahoo MCP servers: 7 ways to get your Wahoo data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/wahoo-mcp-servers
Seven ways to connect Wahoo to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with prices and which ones need Wahoo API approval.
Your Wahoo ELEMNT, KICKR, and TICKR record every ride and workout, and the AI assistant you talk to every day cannot see any of it. The Model Context Protocol (MCP) is the open standard that connects the two, and seven options now do it for Wahoo. This list ranks them by setup effort, from least to most. freddy is ours, and it is first: nothing to install, no API access to apply for, and your rides next to your sleep and recovery in any AI app that speaks MCP. Each entry says who it suits. What to know first Wahoo gives API access only on request, and a live app needs Wahoo's review ( Wahoo developer portal ). So the question that sorts Wahoo options is whether a tool uses a Wahoo app that Wahoo already approved, or asks you to apply yourself. Wahoo also syncs rides to partner platforms, which opens an indirect route: Wahoo to Intervals.icu, then an Intervals.icu MCP server. Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. The list 1. freddy freddy is a hosted MCP server that brings your health data from all your apps into one place. You sign in with a code we email you, connect Wahoo on Wahoo's own sign-in screen, and add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. freddy connects to Wahoo directly, so there is nothing to apply for. New rides and workouts arrive by push soon after your device syncs, and your history comes with them. The same account reads Garmin, Oura, Polar, Apple Health, and more. Our Wahoo guide walks through every step. Setup: three steps in your browser. Nothing to install, nothing to update, and it works on your phone too. Price: free for one source with a short history window, or $49/year for all your sources and your full history. Best for: almost everyone who rides with Wahoo, and especially people who use more than one app. 2. AthleteData AthleteData is a hosted AI coach with a cheaper MCP-only plan. It signs in to Wahoo, and your ELEMNT rides and TICKR workouts sync on their own, next to more than 30 other apps. The coaching plan messages you after workouts; the MCP-only plan leaves that out. Setup: start the trial, connect Wahoo, and add the connector to your AI app. Price: MCP-only is CHF 9 a month or CHF 69 a year, after a 7-day free trial ( plans ). Coaching is CHF 29.90 a month. Best for: athletes who want a coach that gets in touch first, as well as answers. 3. Shape Shape is an endurance training planner with its own MCP server. It signs in to Wahoo with Wahoo's official screen, and it writes too: it sends planned workouts to your ELEMNT. Setup: sign up, connect Wahoo, and add its connector to your AI app. Price: $5 a month, after a 14-day trial ( pricing ). Best for: riders who want the AI to plan sessions that land on their ELEMNT. 4. Tredict Tredict is an endurance training platform with its own MCP server. It connects with Wahoo's sign-in and imports up to 2,000 past activities. It works best if Tredict is already your training log, and its MCP server needs Tredict's paid access for detailed data. Setup: create a Tredict account, connect Wahoo, then add Tredict to your AI app. Price: 2 months free, then $49 for 12 months, paid once (€49 in Europe; prices ). Best for: endurance athletes who already plan their training in Tredict. 5. IcuSync, through Intervals.icu An indirect route. Wahoo syncs rides to Intervals.icu , which is free, and IcuSync serves Intervals.icu over MCP, with tools that read and plan, in Claude on the web, desktop, and phone, and in ChatGPT and Grok. Setup: create an Intervals.icu account, link Wahoo there, sign up for IcuSync, and paste its URL into your AI app. Price: Intervals.icu is free; IcuSync is $20 a year. Best for: athletes who already train in Intervals.icu. 6. Intervals.icu with a local MCP server The do-it-yourself version of the same route: Wahoo syncs to Intervals.icu, and hhopke/intervals-icu-mcp reads it from your own computer with 67 tools, MIT licensed. Intervals.icu's personal API key is free and needs no approval. Setup: moderate. Link Wahoo to Intervals.icu, install uv, copy your athlete ID and API key, and add one command to your AI app's config. Desktop apps only. Price: free and open source. Best for: people at home in a terminal who want every tool on their own computer. 7. armonge/wahoo-mcp A Python server on Wahoo's official API with 9 tools , under GPL-3.0. It needs your own Wahoo developer app, which means applying to Wahoo, and its code last changed in July 2025. Setup: the most work on this list. Apply for Wahoo API access and wait, register an app, install Python 3.13, sign in, and add it to your desktop AI app. Price: free and open source. Best for: developers who already have Wahoo API access. Wahoo MCP options at a glance (prices as of September 2026) Tool Setup Wahoo access Price freddy None; web, desktop, and phone Official Free tier; $49/year AthleteData Account; web, desktop, and phone Official sign-in CHF 69/yr or CHF 9/mo Shape Account; web, desktop, and phone Official sign-in $5/mo Tredict Tredict account; web, desktop, and phone Official sign-in 2 months free, then $49 per 12 months IcuSync Two accounts; web, desktop, and phone Through Intervals.icu $20/yr hhopke/intervals-icu-mcp Terminal and config file; desktop only Through Intervals.icu Free armonge/wahoo-mcp Your own API approval; desktop only Official, your own app Free How to choose For most people, the answer is freddy: nothing to apply for, nothing to run, and your rides next to your recovery data in any AI app that speaks MCP. The exceptions: You want the AI to put workouts on your ELEMNT: Shape. You want a coach that gets in touch first: AthleteData. You already train in Intervals.icu: IcuSync, or hhopke/intervals-icu-mcp on your own computer. Prices and details verified in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Withings MCP servers: 9 ways to get your Withings data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/withings-mcp-servers
Nine ways to connect Withings to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with prices and which ones need your own developer app.
Your Withings scale, blood pressure monitor, watch, and sleep analyzer record a lot about you, and the AI assistant you talk to every day cannot see any of it. The Model Context Protocol (MCP) is the open standard that connects the two, and nine options now do it for Withings. This list ranks them by setup effort, from least to most. freddy is ours, and it is first: nothing to install, no developer app to register, and Withings next to your other apps in any AI app that speaks MCP. Each entry says who it suits. What to know first Withings' API is open to individuals: anyone can register a developer app for free, with no contract ( Withings developer guide ). So the question that sorts the options is whether you just sign in with Withings, or register your own app first. Two details matter if you register your own. The free plan leaves out Withings' paid pack of advanced measures, such as sleep score, overnight HRV, ECG, and VO2 max ( data list ). And an app whose callback is a local address runs in a restricted mode for up to 10 users, which is enough for one person. Withings' own AI assistant lives in the Withings app for Withings+ members, at $9.95 a month or $99.50 a year. Withings ships no MCP server of its own. Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. The list 1. freddy freddy is a hosted MCP server that brings your health data from all your apps into one place. You sign in with a code we email you, connect Withings on Withings' own sign-in screen, and add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. New weigh-ins, blood pressure readings, and nights arrive by push soon after your device syncs, and your history comes with them. The same account reads Garmin, Oura, Polar, Apple Health, and more, so a question like "does my blood pressure follow my training load" works across all of them. Our Withings guide walks through every step. Setup: three steps in your browser. Nothing to install, nothing to update, and it works on your phone too. Price: free for one source with a short history window, or $49/year for all your sources and your full history. Best for: almost everyone with Withings devices, and especially people who use more than one app. 2. withings-mcp.com withings-mcp.com is a free hosted server for Withings alone, open source under MIT. It signs in with Withings' own screen through the project's app and offers 13 read-only tools, from sleep and measurements to ECG and stethoscope recordings. It is a one-person project with no guarantee, and it does not say which Withings plan it uses, which decides whether the advanced measures come through. Setup: add its URL as a custom connector, click Connect, and sign in to Withings. Price: free, with a Patreon for donations. Best for: Withings-only users who want a free hosted server. 3. Wellness Project Wellness Project is a hosted coaching app with its own MCP server. It connects Withings directly for body, sleep, activity, heart, and blood pressure data, and its server lists 73 tools, some of which write to Wellness Project. The free plan allows 3 analysis chats a day, with the app and your AI app counted together. Setup: sign in with Apple or Google, connect Withings in the app, then add its MCP address to your AI app and sign in. Price: a free Basic plan. Pro is $9.99 a month or $99.99 a year, with an introductory $4.99 a month or $39.99 a year until October 15 ( pricing ). Best for: people who want a coaching app alongside their AI assistant. 4. FitnessPeak FitnessPeak , from a Norwegian company, runs a hosted MCP server that covers Withings, WHOOP, and Hevy, with 9 read-only Withings tools for weight, blood pressure, heart, sleep, activity, and more. It connects with a key, so it suits Claude, Cursor, and VS Code. Setup: sign up with a card, connect Withings, and add its key to your AI app. Price: $39 a year, after a 14-day trial ( pricing ). Best for: people who want Withings, WHOOP, and Hevy in one small hosted server. 5. AthleteData AthleteData is a hosted AI coach with a cheaper MCP-only plan. It reads Withings weight, sleep, blood pressure, and body composition, next to more than 30 other apps. The coaching plan messages you after workouts; the MCP-only plan leaves that out. Setup: start the trial, connect Withings, and add the connector to your AI app. Price: MCP-only is CHF 9 a month or CHF 69 a year, after a 7-day free trial ( plans ). Coaching is CHF 29.90 a month. Best for: athletes who want a coach that gets in touch first, as well as answers. 6. Fulcra Context, through Apple Health Fulcra Context is a personal data store with a hosted MCP server. Its iPhone app reads Apple Health, and the Withings app can share weight, blood pressure, heart rate, HRV, sleep, ECG, and more there, which is one way to reach measures that a free-plan API app does not get. It is iPhone only, and the Withings app pushes to Apple Health when you open it. Setup: moderate. Turn on the Apple Health export in the Withings app, install Context, allow Apple Health access, then add Fulcra's server to your AI app. Price: Personal is free. Pro is $14.99 a month or $149.99 a year ( pricing ). Best for: iPhone users who want a broader personal data store. 7. withings-mcp-unofficial A local server with about 18 tools , daily and weekly summaries among them, MIT licensed and published on npm as withings-mcp-unofficial. It reads Withings through your own developer app. Setup: involved. Register your own Withings app with a local callback, install Node, and run its setup, auth, and doctor commands. Desktop apps only. Price: free and open source. Best for: people who want a guided command-line setup on their own computer. 8. partymola/withings-mcp A Python server with 8 tools , a local SQLite cache, and trends, under GPL-3.0. It reads Withings through your own developer app. Setup: involved. Register your own Withings app, install Python 3.13, run its auth command, and add it to your desktop AI app. Price: free and open source. Best for: developers who want a local cache and trends. 9. Open Wearables An open-source platform that you run on your own server, with 2,562 stars and an MIT license. It connects Withings with your own developer app and a public web address for the callback, and it includes an MCP server with 6 tools. Setup: the most work on this list. Run its Docker stack on a server with a public address, register your Withings app, add its keys, and set up the MCP server. Desktop apps only. Price: free software. You pay for the server. Best for: developers who want several sources on hardware they control. Withings MCP options at a glance (prices as of September 2026) Tool Setup Withings access Price freddy None; web, desktop, and phone Official sign-in Free tier; $49/year withings-mcp.com None; web, desktop, and phone Official sign-in Free Wellness Project App and connector; web, desktop, and phone Official sign-in Free (3 chats a day); Pro $9.99/mo FitnessPeak Key; desktop apps and claude.ai Official sign-in $39/yr AthleteData Account; web, desktop, and phone Official sign-in CHF 69/yr or CHF 9/mo Fulcra Context iPhone app; web, desktop, and phone Through Apple Health Free; Pro $14.99/mo withings-mcp-unofficial Your own Withings app; desktop only Official, your own app Free partymola/withings-mcp Your own Withings app; desktop only Official, your own app Free Open Wearables Your own server and Withings app; desktop only Official, your own app Free; you pay for hosting How to choose For most people, the answer is freddy: no developer app, nothing to keep running, and Withings next to your other apps in any AI app that speaks MCP, on your phone too. The exceptions: You use only Withings and want it free: withings-mcp.com. You want a coaching app alongside: Wellness Project. You want Withings, WHOOP, and Hevy in one small server: FitnessPeak. You want it on your own computer: withings-mcp-unofficial. Prices and details verified in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Apple Health MCP servers: 10 ways to get your Apple Health data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/apple-health-mcp-servers
Ten ways to connect Apple Health to Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with how fresh the data is and a price for each one.
Apple Health collects your steps, sleep, heart rate, and workouts from your iPhone, your Apple Watch, and many other apps, and the AI assistant you talk to every day cannot see any of it. The Model Context Protocol (MCP) is the open standard that connects the two, and at least ten tools now do it for Apple Health. They differ most in how fresh the data is and how much work they take. This list ranks them by setup effort, from least to most. freddy is ours, and it is first: one iPhone app, nothing to export, nothing to keep open, and it works in any AI app that speaks MCP. Each entry says who it suits. What to know first Apple offers no cloud API for Health data, and it ships no MCP server. Your data leaves the iPhone in one of two ways: the Health app's "Export All Health Data" file, or an iPhone app that you allow to read Apple Health. A Mac cannot read Apple Health at all ( Apple's documentation ), so every Mac tool here gets its data from an iPhone app or an export file. That makes freshness the real question. An export is a snapshot: it is as old as the day you made it, and you repeat it for new data. An app with Apple Health access can send new data as it arrives, within the limits iOS sets on background work. A few tools read the phone live, but only while their app is open. Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. Claude and ChatGPT can also read Apple Health without MCP. Claude's iPhone app does it in beta, for Pro and Max plans in the US ( Claude help ). ChatGPT Health syncs Apple Health from an iPhone for Free, Go, Plus, and Pro users in the US. Each works inside its own assistant only. Outside the US, or in more than one AI app, you need one of the options below. The list 1. freddy freddy is a hosted MCP server for your health data from all your apps. For Apple Health you install freddy for iPhone , allow it to read Apple Health, and add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. freddy reads your past data as well as new data. Your iPhone sends new data in the background when iOS schedules it, and right away when you open freddy, so the data is recent without an export or an open app. Because freddy is hosted, you can ask from your laptop or your phone, wherever your iPhone is. freddy also connects Garmin, Oura, Polar, Withings, and more directly, so you can ask across all of them. Our setup guide walks through every step. Setup: install the iPhone app, allow Apple Health access, and add freddy to your AI app. No computer, no export files. Price: free for one source with a short history window, or $49/year for all your sources and your full history. Best for: almost everyone with an iPhone, and especially people who use more than one app or want to ask from any device. 2. Fulcra Context Fulcra Context is a personal data store with a hosted MCP server. Its iPhone app uploads Apple Health, and after the first sync it keeps syncing in the background. You sign in to the server with OAuth, so it works in Claude on the web, desktop, and phone, and in ChatGPT on the web. It is a general personal data store rather than a health app, and its docs do not say how often it syncs. Setup: install the Context app, create an account, allow Apple Health access, wait for the first sync, then add Fulcra's server address to your AI app and sign in. Price: Personal is free and includes the MCP server. Pro is $14.99 a month or $149.99 a year ( pricing ). Best for: people who want a free hosted server for Apple Health alone. 3. Wellness Project Wellness Project is an AI coaching app for iPhone with its own MCP server. The app imports Apple Health and keeps the synced history, and its server offers 73 tools, some of which write to Wellness Project. It works as a connector in Claude, ChatGPT, Gemini, and Grok. The free plan allows 3 analysis chats a day, with the app and your AI app counted together. Its page does not say how often it syncs. Setup: install the app, create an account, turn on Apple Health in its settings, then add its MCP server to your AI app and sign in. Price: a free Basic plan. Pro is $9.99 a month or $99.99 a year, with an introductory $4.99 a month or $39.99 a year until October 15 ( pricing ). Best for: people who want a coaching app alongside their AI assistant. 4. Shape Shape is an endurance training planner with its own MCP server. Its iPhone app reads Apple Health: workouts first, plus steps, sleep, HRV, resting heart rate, VO2 max, and body composition. The server can also change your training plan. It is a training tool first. For all of Apple Health, look at entries 1 to 3. Setup: sign up, install the iPhone app, allow Apple Health access, and add the connector to your AI app. In ChatGPT it needs a paid plan. Price: $5 a month, or $59.99 once for lifetime access, after a 14-day trial ( pricing ). Best for: endurance athletes who plan their training with AI. 5. Health Auto Export Health Auto Export runs an MCP server on the iPhone itself. Your AI app asks the phone, and the phone reads Apple Health at that moment, so the data is as fresh as it gets. The cost of that freshness: the app must stay open on screen while you ask, and your computer must be on the same Wi-Fi. Because the server lives on your home network, it works with desktop apps such as Claude Desktop, Cursor, and Claude Code, and not with claude.ai, phone apps, or ChatGPT. Setup: moderate. Buy Premium, start the server in the app, copy its address and token, and add them to your desktop AI app. Keep the app open each time you use it. Price: Premium is $1.99 a month, $6.99 a year, or $24.99 once, after a 7-day trial. The Basic plan does not include the server ( App Store ). Best for: people who want live data at their desk and do not mind keeping the app open. 6. MetricBridge with health-export-mcp The MetricBridge iPhone app exports 190 metrics as JSON files to iCloud Drive, and refreshes them in the background when new data arrives. The open-source health-export-mcp server on your computer reads those files, with 14 read-only tools under an MIT license. Freshness depends on iOS background time and on iCloud, which may need a few minutes to reach your computer. Setup: moderate. Install the app, allow Apple Health access, choose iCloud, then add the server to your desktop AI app with the folder path. Desktop apps only. Price: the app is $24.99 once, after a 7-day trial. The server is free. Best for: people who want plain files they can read, and a free local server. 7. Health.md Health.md exports Apple Health to Markdown for Obsidian, and pairs a Mac app with its iPhone app to offer a local MCP server with 21 tools, under AGPL-3.0. The iPhone app must stay open while the Mac asks, and both must be on the same network. A cross-platform command-line version is in early preview. Setup: moderate. Install both apps, turn on Mac connectivity, pair them, and add the helper to your desktop AI app. Desktop apps only. Price: 10 free exports, then $19.99 once for full access. Best for: Obsidian users with a Mac. 8. neiltron/apple-health-mcp The most-starred Apple Health MCP server, with 568 stars and an MIT license. It loads CSV files into DuckDB and gives your AI app 3 tools to query them. It does not read the Health app's own export. It needs CSV files from the Simple Health Export CSV app, which was last updated in 2022, and your data is as old as your last export. Setup: involved. Export from that app, move the files to your computer, unzip them, and add the server with the folder path. Repeat the export for new data. Desktop apps only. Price: free and open source. Best for: people who want SQL over a snapshot of their data. 9. vpetersson/apple-health-mcp-server A Rust server that reads the Health app's own export: 12 read-only tools over your records, workouts, ECGs, and routes, stored in DuckDB, MIT licensed and updated this week. Its README describes Homebrew and one-click installs, but no release exists yet, so today you build it from source. Your data is as old as your last export, and a new import skips duplicates. Setup: involved. Export from the Health app, copy the ZIP to your computer, build the server with Rust, run the import, and add it to your desktop AI app. Price: free and open source. Best for: developers who want the native export and no third-party iPhone app. 10. Open Wearables An open-source platform that you run on your own server, with 2,558 stars and an MIT license. For Apple Health it takes an export upload, or a beta iPhone app that syncs in the background, by invitation through its Discord. It includes an MCP server with 6 tools and covers several other wearables. Setup: the most work on this list. Run its Docker stack, create an API key, set up the MCP server, then upload your export or join the beta app. Desktop apps only. Price: free software. You pay for the server. Best for: developers who want several sources on hardware they control. Apple Health MCP options at a glance (prices as of September 2026) Tool Data freshness Setup Price freddy Background sync, and on app open iPhone app; web, desktop, and phone Free tier; $49/year Fulcra Context Background sync iPhone app; web, desktop, and phone Free; Pro $14.99/mo Wellness Project Sync, interval not stated iPhone app; web, desktop, and phone Free (3 chats a day); Pro $9.99/mo Shape Sync, interval not stated iPhone app; web, desktop, and phone $5/mo or $59.99 once Health Auto Export Live, while the app is open iPhone app on the same Wi-Fi; desktop only $1.99/mo, $6.99/yr, or $24.99 once MetricBridge Background file export through iCloud iPhone app and local server; desktop only $24.99 once Health.md Live, while the app is open iPhone app and Mac app; desktop only $19.99 once neiltron Last manual export Export app and local server; desktop only Free vpetersson Last manual export Build from source; desktop only Free Open Wearables Export upload or beta app Your own server; desktop only Free; you pay for hosting How to choose For most people, the answer is freddy: one iPhone app, nothing to export, nothing to keep open, and it works in any AI app that speaks MCP, on your phone too. The exceptions: You want a free hosted server for Apple Health alone: Fulcra Context. You want a coaching app alongside: Wellness Project, or Shape for endurance training plans. You need data live to the minute at your desk: Health Auto Export. You want files and code you control: MetricBridge with health-export-mcp, or vpetersson's server for the native export. You want several sources on hardware you control: Open Wearables. Prices and details verified in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Fitbit and Google Health MCP servers: 8 ways to get your data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/fitbit-mcp-servers
Eight ways to connect your Fitbit, now part of Google Health, to Claude, ChatGPT, or any AI app over MCP now that the old Fitbit Web API is closing. Ranked by setup effort, with prices.
Your Fitbit holds years of your steps, sleep, and heart rate, and the AI assistant you talk to every day cannot see any of it. The Model Context Protocol (MCP) is the open standard that connects the two. For Fitbit there is a catch: Google turns off the Fitbit Web API that most Fitbit MCP servers were built on, on November 2, 2026. This list covers only the options that still work after that. freddy is ours and sits first: no Google Cloud project, nothing to keep running, and it works in any AI app that speaks MCP. After it, the list runs from least setup effort to most. Each entry says who it suits. What to know first Google ended support for the old Fitbit Web API on September 30, 2026, and turns it off on November 2, 2026 ( Google Health API ). Its replacement is the Google Health API, and the Fitbit app itself is now called Google Health. A Fitbit MCP server that asks you to register an app at dev.fitbit.com is built on the old API, so skip it: it stops working on November 2. Old Fitbit sign-ins are gone too, so the Google Health app now needs a Google account ( Google's account guide ). That leaves two ways in. The first is the Google Health API, which signs you in with your Google account. A hosted service handles that for you, while a server you run yourself needs its own Google Cloud project, which is real work. The second is your phone: the Google Health app can share your data into Health Connect on Android and Apple Health on iPhone, and other apps read it from there. The phone route has gaps, because Google does not share every metric. On Android it leaves out blood oxygen (SpO2), and on iPhone it leaves out HRV and skin temperature ( Android list , iPhone list ). One trap for every server you run yourself: while your Google Cloud app is in Testing, Google ends the sign-in after 7 days. Publish the app in the Google Cloud console, then sign in again. Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. Claude and ChatGPT can also read phone health data without MCP. Claude's phone apps read Apple Health and Health Connect, in beta for Pro and Max plans in the US. ChatGPT Health reads Apple Health from an iPhone, in the US. Each works inside its own assistant only, and your Fitbit data reaches them only through the Google Health share. Outside the US, or in more than one AI app, you need one of the options below. The list 1. freddy freddy is a hosted MCP server for your health data from all your apps. For Fitbit it uses the phone route: the Google Health app shares into Health Connect or Apple Health, and freddy's phone app reads it from there. In freddy, pick Fitbit, and it shows you where to turn on sharing in the Google Health app. Then add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. There is no Google Cloud project to set up. The same account reads Garmin, Oura, Polar, Withings, and more, so you can ask across all of them. Your phone sends new data in the background when the system allows it, and right away when you open freddy. There is also a direct Google Health connection in alpha. The limits come from the phone route: HRV reaches freddy only from Android, and blood oxygen only from iPhone. Our setup guide walks through every step. Setup: install freddy on your phone, connect Apple Health or Health Connect, turn on Google Health's sharing, and add freddy to your AI app. No server and no developer account. Price: free for one source and 7 days of history, or $49/year for all your sources and your full history. Best for: most Fitbit users, and anyone who wants Fitbit next to their other apps in one place. 2. Wellness Project Wellness Project is a hosted coaching app with its own MCP server. It connects Fitbit through the Google Health API, so you sign in with Google and your phone plays no part. The vendor lists 73 tools, some of which write to Wellness Project. The free plan allows 3 analysis chats a day, with the app and your AI app counted together. Adding it in ChatGPT takes a paid plan such as Plus or Pro, because custom connectors there need developer mode. Setup: create an account, connect Fitbit through Google, then add its MCP address to your AI app and sign in. Price: a free Basic plan. Pro is $9.99 a month, and local prices may vary ( pricing ). Best for: people who want Fitbit in their AI app without a phone app, and a coaching app alongside. 3. Fulcra Context Fulcra Context is a personal data platform with a hosted MCP server. Its iPhone app reads Apple Health, so Fitbit data arrives once you turn on the Google Health share to Apple Health. Fulcra's pages do not mention Fitbit; the route works because Fulcra reads whatever Apple Health holds. It is iPhone only (iOS 17 or later), and through this route it gets no HRV from Fitbit. Setup: turn on the Apple Health share in the Google Health app, install Context, sign up, then add Fulcra's server to your AI app. Price: Personal is free. Pro is $14.99 a month or $149.99 a year ( pricing ). Best for: iPhone users who want a broader personal data store, not only health. 4. gorban-dev/google-health-mcp A local server with a setup wizard: 20 tools , 7 of which write food, water, and weight logs, down to logging a meal from a photo. MIT licensed. Setup: involved, on your own computer. Run its setup command in a terminal, and the wizard takes you through your own Google Cloud project and OAuth client, including the step that stops the 7-day cutoff. Desktop apps only. Price: free and open source. Best for: people at home in a terminal who want the gentlest self-run setup. 5. partymola/google-health-mcp 19 tools with a local SQLite cache, trend tools, and ECG readings, under GPL-3.0. It needs Python 3.13, and its guide to the Google Cloud steps is the clearest of the self-run options. Setup: involved. Seven steps in the Google Cloud console, then one command to add it to Claude Code, which its docs target. Desktop apps only. Price: free and open source. Best for: developers who use Claude Code and want trends and a local cache. 6. davidmosiah/google-health-mcp The most-starred server on the new API, with 74 stars , an MIT license, a beta label, and 26 tools that include setup and diagnostic helpers. Setup: involved. Set up a Google Cloud OAuth client with the redirect address from its README, then run its setup, auth, and doctor commands. The README does not mention the 7-day cutoff, so publish your Google app before you sign in. Desktop apps only. Price: free and open source. Best for: developers who want the most-used option and its diagnostic tools. 7. akshaygoyal/google-health-mcp A server you deploy yourself on Cloudflare Workers: 28 read-only tools for 38 data types, MIT licensed. Because it runs on the web, it works in claude.ai and on your phone, which the local servers cannot do. Setup: the most of the servers here. You need a Google Cloud OAuth client, a Cloudflare account, Cloudflare's wrangler tool for storage and secrets, a token script, and a deploy. Then you paste a secret URL into your AI app. Price: free, within Cloudflare's free tier. Best for: developers who want a server they own that also works on their phone. 8. Open Wearables An open-source platform that you run on your own server, with 2,717 stars and an MIT license. Since version 0.9.0 it has separate Google Health and Health Connect providers, and it includes an MCP server with 6 tools. It covers several other wearables too. Setup: the most work on this list. You run a Docker stack with a database, set up a Google Cloud OAuth client and its keys, create an API key, and set up the MCP server. Desktop apps only. Price: free software. You pay for the server. Best for: developers who want several sources on hardware they control. Fitbit MCP options at a glance (prices checked on 26 October 2026) Tool Setup Fitbit route Price freddy Phone app; web, desktop, and phone Phone share (Health Connect or Apple Health) Free tier; $49/year Wellness Project None; web, desktop, and phone Google Health API Free (3 chats a day); Pro $9.99/mo Fulcra Context iPhone app; web, desktop, and phone Phone share (Apple Health) Free; Pro $14.99/mo gorban-dev Terminal and Google Cloud project; desktop only Google Health API, your own project Free partymola Terminal and Google Cloud project; desktop only Google Health API, your own project Free davidmosiah Terminal and Google Cloud project; desktop only Google Health API, your own project Free akshaygoyal Cloudflare deploy and Google Cloud project; web, desktop, and phone Google Health API, your own project Free (Cloudflare free tier) Open Wearables Your own server and Google Cloud project; desktop only Google Health API or Health Connect Free; you pay for hosting How to choose For most people, the answer is freddy: no Google Cloud project, nothing to keep running, and it works in any AI app that speaks MCP, on your phone too. The exceptions: You want Fitbit in your AI app without a phone app: Wellness Project. You want a broad personal data store on your iPhone: Fulcra Context. You want to run it yourself: gorban-dev/google-health-mcp for the gentlest setup, or akshaygoyal/google-health-mcp if it must work on your phone. You want several sources on hardware you control: Open Wearables. Whatever you pick, skip any Fitbit MCP server that asks for a dev.fitbit.com app. It is built on the API that Google is turning off. Prices, star counts, and Google's dates were checked on 26 October 2026. Other third-party details were checked in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## Oura MCP servers: 8 ways to get your Oura data into Claude, ChatGPT, or any AI client
URL: https://freddy.coach/articles/oura-mcp-servers
Eight third-party ways to link your Oura Ring with Claude, ChatGPT, or any AI app over MCP, ranked by setup effort, with prices.
Your Oura Ring tracks your sleep, readiness, and heart rate every night, and the AI assistant you talk to every day cannot see any of it. The Model Context Protocol (MCP) is the open standard that connects the two, and eight options now do it for Oura. Many older guides no longer work, so this list shows only the ones that do. This list covers eight third-party options for Oura. It ranks them by setup effort, from least to most. freddy is ours, and it is first: nothing to install, no developer account, and it works in any AI app that speaks MCP. Each entry says who it suits. What to know first Oura also operates its own MCP server. The options below are third-party routes. Most options sign in with OAuth. A hosted service does that through its own Oura app. A server you run yourself needs an Oura developer app of your own, which you register at Oura's developer portal . One more rule applies to every option on this list: if your ring is Gen3 or newer, Oura requires an active Oura Membership for API access, including through partner apps ( Oura help ). Where a server runs decides where you can use it. A hosted server is a web address, so it works in Claude and ChatGPT on the web, on the desktop, and on your phone. A server on your computer works only in desktop apps such as Claude Desktop and Cursor. The Claude phone app and claude.ai cannot reach it, and ChatGPT needs a tunnel you set up yourself. ChatGPT and Claude can read some Oura data without MCP, through Apple Health on an iPhone, in the US: ChatGPT Health on Free to Pro plans, and Claude in beta on Pro and Max. Oura shares only part of its data with Apple Health, such as sleep stages, heart rate, steps, and workouts, and not readiness, sleep score, HRV, or temperature ( Oura help ). The list 1. freddy freddy is a hosted MCP server for your health data from all your apps. You sign in with a code we email you, connect Oura on Oura's own sign-in screen, and add freddy to your AI app, from Claude's connector directory or ChatGPT's Plugin Directory , or with one personal URL in Cursor or any other MCP client. freddy uses Oura's official API with OAuth, and you do not register a developer app. New Oura data arrives soon after your ring syncs, and freddy reads your history too, readiness and sleep scores included. The same account reads Garmin, Polar, Withings, Apple Health, and more, so you can ask across all of them. Like every option that uses Oura's API, freddy needs an active Oura Membership for a Gen3 or newer ring. Our Oura guide walks through every step. Setup: three steps in your browser. Nothing to install, nothing to update, and it works on your phone too. Price: Free is $0 for 1 source and 7 days of history shown. Pro is $49/year for all your sources and your full history, and Believer is $249 once for life. Best for: almost everyone, and especially people who use more than one app or would rather not run a server. 2. Wellness Project Wellness Project is a hosted coaching app with its own MCP server. It connects Oura through its own Oura app, so you only approve access on Oura's screen, and it offers the synced history to your AI app. Its server lists 73 tools, some of which write to Wellness Project. The free plan allows 3 analysis chats a day, with the app and your AI app counted together. Setup: sign in with Apple or Google, connect Oura, then add its MCP address to your AI app and sign in. Price: $0 for Basic. Pro is US$9.99 a month. Local prices may vary ( pricing ). Best for: people who want a coaching app alongside their AI assistant. 3. AthleteData AthleteData is a hosted AI coach with a cheaper MCP-only plan. It connects Oura with one sign-in, and it lists 17 Oura tools and about 180 days of starting history. The coaching plan messages you after workouts; the MCP-only plan leaves that out. Setup: sign up, connect Oura, copy your personal MCP address, and add it to your AI app. Price: MCP-only is US$9 a month or US$69 billed annually. Coaching is US$39 a month. Try either plan free for 7 days ( USD prices ). Best for: athletes who want a training coach as well as answers. 4. Rajskij/oura-mcp A small local server with a one-click install for Claude Desktop: 10 read-only tools , MIT licensed, with a demo mode on Oura's sample data that needs no account. Setup: moderate. Register your own Oura app, open the .mcpb file in Claude Desktop, paste your app's client ID and secret, and approve access in the browser. A self-hosted web version exists for claude.ai and phones, with more setup. Price: $0 for the software. Best for: Claude Desktop users who want the quickest self-run setup. 5. davidmosiah/oura-mcp About 23 tools , published on npm as oura-mcp-unofficial under an MIT license, with setup, auth, and doctor commands that write your Claude Desktop config for you. Setup: moderate. Install Node 20, register your own Oura app, run the setup and auth commands, and let it add itself to your AI app. Desktop apps only. Price: $0 for the software. Best for: people who want a guided command-line setup. 6. mitchhankins01/oura-ring-mcp The most-starred working option, with 29 stars and 17 tools: 10 for your data and 7 that analyze it, plus prompts. Use the OAuth setup its README lists. The README says MIT, but the repository has no license file, and its last change was in February 2026. Setup: moderate. Install Node 18, register your own Oura app, run its auth command, and add your app's ID and secret to Claude Desktop's config. Desktop apps only, unless you deploy it yourself. Price: $0 for the source code. Best for: people who want built-in analysis tools. 7. Open Wearables An open-source platform that you run on your own server, with 2,558 stars and an MIT license. It signs in to Oura with your own app, receives Oura's update notifications, and includes an MCP server with 6 tools. The project calls itself early-stage. Setup: the most work of the ready-made options. Run its Docker stack, put your own Oura app's keys in its config, create an API key, and set up the MCP server. Oura's notifications need a public web address. Desktop apps only. Price: $0 for the software. You pay for the server. Best for: developers who want several sources on hardware they control. 8. Oura's API spec with an OpenAPI-to-MCP tool Oura publishes an OpenAPI spec , so a generic tool such as openapi-mcp-generator can turn it into an MCP server with a tool for each endpoint. The sign-in is yours to solve: the generator cannot run Oura's OAuth flow, so you get a token yourself and pass it in. Setup: the most work on this list. Register your own Oura app, generate the server, get a token, and renew it by hand. Desktop apps only, unless you host it. Price: $0 for the software. Best for: developers who want full control over every tool. Oura MCP options at a glance (prices checked on 25 October 2026) Tool Setup Oura sign-in Price freddy None; web, desktop, and phone Official OAuth $0 tier; $49/year Wellness Project None; web, desktop, and phone Official OAuth $0 (3 chats a day); Pro US$9.99/mo; local prices vary AthleteData Connector address; web, desktop, and phone Official OAuth US$69/yr or US$9/mo Rajskij/oura-mcp One-click file and your own Oura app; Claude Desktop OAuth, your own app $0 davidmosiah/oura-mcp Terminal and your own Oura app; desktop only OAuth, your own app $0 mitchhankins01 Terminal and your own Oura app; desktop only OAuth, your own app $0 Open Wearables Your own server and Oura app; desktop only OAuth, your own app $0 software; you pay for hosting OpenAPI generator Generate, then renew tokens by hand Token you get yourself $0 How to choose For most people, the answer is freddy: no developer app, nothing to keep running, and it works in any AI app that speaks MCP, on your phone too. The exceptions: You want a coaching app alongside: Wellness Project, or AthleteData for training. You use Claude Desktop and want to run it yourself: Rajskij/oura-mcp. You want several sources on hardware you control: Open Wearables. Whatever you pick, choose one that signs in with OAuth. Prices are in US dollars and were checked on 25 October 2026. Other third-party details were checked in September 2026. This space moves fast, so check each vendor's site for current details before you commit. For other wearables, see our roundup of wearable MCP servers .

## What's new in freddy: week of September 14, 2026
URL: https://freddy.coach/articles/whats-new-2026-09-20
More languages for web account settings, guided setup, searchable help, fixes for Hevy and Garmin, and smaller bugs squashed.
A warm welcome to the 3,299 new members who joined freddy last week. Here’s what changed: In short Web account settings and navigation in more languages. Clearer guided setup for sources and AI connections. Search and topic browsing in the help center. More reliable Hevy workout history. More complete Garmin summaries when activity types overlap. Various bugs and rough edges squashed across connections, imports and setup. Web settings in more languages Account settings and navigation on the web now support German, Spanish, French, Italian, Japanese, Korean, Polish and Simplified Chinese, alongside English. Choose a language in Settings or follow your browser’s language. This covers account controls and navigation. Source setup, data views, support, emails and help articles remain in English. The translated controls include language selection, account and billing settings, notifications, logged-in devices, data exports and account deletion. Your choice is saved to your account. These translations apply to the signed-in web controls described here, rather than the whole product. Guided web setup Adding a source on the web now follows a focused flow with the relevant connection steps and a way to return to unfinished setup. AI connections also have step-by-step guidance. Completed setup no longer leaves a recurring completion notice on Overview. Source and AI setup now keep their connection steps together, and unfinished source setup stays reachable when you return. Find help by source or topic The help center now has article search, topic navigation and direct starting points for connecting a source, connecting an AI client or finding missing data. You can look for the guide you need without scanning the full article list. Search is useful when you know the source name or the problem. Topic browsing gives you another route when you are still working out which guide applies. The existing guides remain available from their direct links. More reliable Hevy history Hevy history imports now work through your saved workout list, including older sessions that could be missed. Later edits and deletions are also reconciled with your Hevy log. Hevy requires a Hevy Pro subscription for its connection. More complete Garmin activity summaries When Garmin reports different activity types for the same period, freddy now combines their detail instead of letting one type replace another. The fix applies to newly received activity summaries. Various bugs and rough edges squashed A round of smaller fixes landed across source connections, history imports and setup. They make everyday flows more reliable without changing how you use freddy.

## What's new in freddy: week of September 7, 2026
URL: https://freddy.coach/articles/whats-new-2026-09-13
Amazfit/Zepp beta for all Pro users, redesigned source cards, bug fixes and performance improvements, and Google Health alpha.
In short Amazfit/Zepp is in beta, available to all Pro users. Redesigned source cards make your connections easier to manage. Lots of bug fixes and performance improvements across freddy. Google Health is now in alpha. Amazfit/Zepp is in beta Amazfit/Zepp is now in beta and available to all Pro users. Connect your Zepp account from Sources on the web dashboard to bring supported sleep, activity and workout data into freddy. Source cards, redesigned Source cards on the web dashboard have been redesigned for easier control. Connection status, sync progress and everyday actions are easier to find, so managing your sources takes less searching. Bugs squashed, performance improved A broad round of bug fixes and performance improvements landed across freddy this week, covering syncing, connections and the dashboard. The focus was on making everyday use smoother and more reliable. Google Health enters alpha Google Health is now in alpha, adding a direct connection to freddy. Access is limited during this early testing phase while the integration is tested and refined.

## What's new in freddy: week of Aug 17, 2026
URL: https://freddy.coach/articles/whats-new-2026-08-23
Suunto workouts now store per-second streams, Garmin activity dynamics import from your data export, and fresh data no longer waits behind large history backfills.
In short Suunto workouts now carry per-second speed, cadence, power, altitude, vertical speed and temperature, plus heart rate beat by beat. Garmin's running and cycling dynamics import from your data export, so ground contact time and vertical oscillation reach your older activities. New data no longer queues behind somebody else's large history import. The dashboard's “Latest data” tile reads your newest data day, and a separate row shows whether a sync is running. Suunto workouts, second by second Until last week a Suunto workout in freddy was a set of summary numbers and, since the week before that, the route you recorded. What was missing was everything that happened between the start and the finish. That is now stored. Speed, cadence, power, altitude, vertical speed and temperature come through at roughly one reading a second, and heart rate comes through beat by beat rather than averaged into a tick. Workouts already synced come back once on their own to collect the streams, so there is nothing to re-import. The practical difference is the kind of question you can ask. A summary tells your AI client that a run averaged a certain pace. The streams let it look at where that pace went, what your cadence did on the climb, and whether your heart rate was already drifting before the split you remember as the hard part. Garmin activity dynamics, going backwards Garmin records running and cycling detail inside the activity files themselves: ground contact time, vertical oscillation, vertical ratio, stride length, left and right balance, pedal smoothness, torque effectiveness. The data export Garmin gives you does contain them. The Garmin source now has an Import activity dynamics button that reads that export ZIP and fills the detail in behind you, using the same file you may already have uploaded for the rest of your history. The ZIP is opened in your browser, and only the running, cycling and swimming activity files are sent. The rest of the archive never leaves your machine. Data stops waiting in line Connecting a source with years of history produces a lot of data at once, and until last week all of it went into a single queue alongside everyone else's live data. If somebody started a large backfill, everything behind it waited. On the worst morning that meant a five hour delay before a night's sleep appeared. Fresh data is back to arriving within seconds, and a large import no longer costs anybody else their morning. A dashboard that agrees with itself The Overview's Latest data tile was reading the last time a batch arrived rather than the newest day of data inside it. For sources that push from your phone the two are very different, so the tile could read “just now” directly above a coverage chart whose final bars were empty. Several people reported that as freddy lying to them, which is a fair description of it. The tile now answers its own question: how old your newest data is, and which source it came from. Whether freddy is live or currently syncing moved to a row of its own, so neither figure has to stand in for the other.

## What's new in freddy: week of Aug 10, 2026
URL: https://freddy.coach/articles/whats-new-2026-08-16
freddy now tells you when a source stops syncing. Plus Suunto GPS tracks, two repairs to Garmin history, and a new recipe.
In short freddy now sends a notification when one of your sources stops syncing, and it names the source. Suunto workouts bring the route you recorded, not just the summary numbers. Garmin history: freddy now explains the permission before you authorize, and offers a Reconnect to sources that missed it. Large Garmin data exports import their full activity history instead of dropping it. A new recipe walks through a triathlon plan that adjusts itself every morning. Claude Code and Codex have their own setup rows, each with one command to copy. Notifications when a source goes quiet A source that stops sending data is the failure mode you are least likely to catch. Nothing breaks, nothing errors, the data just stops, and you find out weeks later when you ask a question and the answer stops short. freddy now watches for that and sends a notification naming the source that went quiet, rather than claiming your whole account went dark. Notifications arrive in the iOS and Android app, and the category can be switched off in Settings or from the dashboard. Suunto workouts carry the recorded track Suunto workouts used to arrive as summary numbers, so a run was a distance and a pace with nothing about where it went. The recorded track now comes through as well, latitude, longitude and altitude across the workout, which makes questions about place answerable: how a repeated route compared week to week, or where the climbs on a session actually landed. Existing workouts fill in as they sync, so older sessions catch up on their own. Garmin history, and the permission that controls it freddy now explains the Garmin history permission before you authorize, on first connect and on reconnect. Sources that missed it show a Reconnect button, and granting the permission re-runs the backfill. Large Garmin data exports keep their activities freddy can import the data export Garmin gives you for older history. Large exports used to lose activities, and now import in full. If an earlier import landed without your activities, re-upload the export and they will come through. A new recipe: a triathlon plan that adjusts itself Hold your training plan in a project, check it each morning against what your body actually did, and move today's session when the plan and reality disagree. The recipe walks through the whole loop, including the part most plans skip: noticing the sessions you quietly did not do. It needs a source for training sessions and one for sleep and recovery, plus a calendar connector if you want the week rewritten for you rather than just proposed. The recipe is at freddy.coach/recipes/adaptive-triathlon-plan. Claude Code and Codex have their own setup Both are now their own rows in the connect-an-AI picker, each with a single command to copy and run instead of an MCP URL to place by hand.

## What's new in freddy: week of Jul 27, 2026
URL: https://freddy.coach/articles/whats-new-2026-08-02
Questions about older dates no longer come back empty, the metric catalogue your AI reads at the start of a session is about half the size, and the Android update is live on Play.
In short Your AI stops reporting an older date as empty when the records were there all along. The catalogue of metrics it reads at the start of a session is about half the size. The Android update is live on Play, with battery guidance for your phone brand. A source stuck mid sync can always be removed now. The support chat has a consent toggle for sync troubleshooting, off unless you turn it on. Apps behind Apple Health and Health Connect answer properly, and records dated years ahead are refused. Older dates stop coming back empty freddy's query tool takes one date range at a time and returns the newest records first. Ask your AI to compare two dates a few weeks apart and it tends to reach for a single wide range bracketing both. If that range holds more records than one response can carry, the oldest end gets cut, which is exactly the date that prompted the question. Your AI then reports that date as empty. A truncated response now names the calendar dates in the part it left out that do hold records, so your AI can see which days are still waiting rather than calling them missing. The tool description also states plainly that two separate dates need two separate calls. A lighter catalogue at the start of every session Every session begins with your AI reading the list of metrics your account holds, because metric names are source specific and cannot be guessed. That list was keyed by metric, unit, source and device, so one metric name fanned out across every watch, phone and app that had ever written it. On a busy Apple Health account it ran to hundreds of lines before you had asked anything. Entries are now collapsed to one per metric and unit per source, with the devices listed inline, grouped under a single heading per source, and paged rather than running on. Every metric name is still there, and more of the conversation is left for your actual question. The Android update is on Play Health Connect background sync used to stop when another app had written a record it could not read. It skips that record and carries on now. The Android app also detects battery settings that block background syncing and shows the steps for your phone brand, since those settings differ a lot between manufacturers. The status telling you your history is still importing now appears on Android too, the same as on iPhone. A stuck source can always be removed A source that was mid sync, or stuck reporting that it was syncing, refused to disconnect. That is the wrong behavior for a failing source, which is precisely when someone wants it gone. Removal now always goes through, and any sync work still running for it is cancelled and cannot write anything back afterwards. A consent toggle for support Working out why a source stopped syncing needs a look at your sync state. The support chat now carries a toggle for that, off unless you switch it on, granting a narrow view of your sync state. Every time that view is opened a line lands in your own chat thread saying so, and switching the toggle back off ends the access. Smaller fixes Apps whose data reaches freddy through Apple Health or Health Connect, such as Samsung Health, Sleep Cycle, MyFitnessPal and Peloton, used to fail with a generic error when you asked your AI to connect or sync them. They now get a real answer naming the hub that carries their data, and it is platform accurate. A source app writing the wrong year could store a record dated years into the future, which then counted as the newest thing in your history, so a question about your latest weight answered from a phantom. Records dated more than a couple of days ahead are refused as they arrive now, for both Apple Health and Health Connect.

## What's new in freddy: week of Jul 20, 2026
URL: https://freddy.coach/articles/whats-new-2026-07-26
Apple Health forward syncing is fixed on iPhone, the Personal API opened in beta, install buttons replace copy-and-paste setup, and 150+ more source apps are recognized.
Four changes stand out from the week of July 20: a fix for Apple Health syncing that had stalled part of the metric catalog on iPhone, a read-only Personal API in beta, install buttons that replace the copy-and-paste setup step, and recognition for well over 150 more apps that write into Apple Health and Health Connect. Apple Health syncing catches back up On iPhones with a lot of history to work through, an incremental sync could spend its entire budget on one heavy metric type and never reach the rest. The visible effect was confusing: steps and heart rate kept arriving, so the connection looked healthy, while most of the catalog quietly stopped moving forward. Each metric type now gets its own share of the work, so no single heavy type can crowd out the others. Accounts that were affected recover on their own over a few days of normal use. There is nothing to re-import and no need to reconnect. Health permission prompts also stopped appearing during background syncs. They now show only while the app is open, so a background sync can no longer sit waiting on a prompt nobody is there to answer. Both fixes are in version 1.0.4 on the App Store. A read-only API for your own data freddy's main path is an MCP connection, which is what your AI client speaks. Not everything is an AI client, though, and the Personal API covers the rest. You create a key in the dashboard and read your own data over plain HTTP: your normalized metrics, the catalog of metrics and units you actually have, your connected sources, and your profile. It is read-only, so a key can see your data but never change it or your account. That makes it a reasonable fit for a script, a home dashboard, or anything else that wants numbers rather than a conversation. The full reference lives at freddy.coach/api-docs. The Personal API is in beta, so access is requested from the dashboard. Installing takes one click Connecting freddy to an AI client used to mean copying your personal MCP URL out of the dashboard and pasting it into the client's connector settings. That step is now a button. Claude gets a deep link that carries the URL across, and ChatGPT opens freddy's app listing. The copy-and-paste path still works for clients that need it, and for anyone who prefers it. Grok also joined the list of AI clients with their own setup instructions, alongside Claude, ChatGPT, Perplexity, and the command-line clients. More of your apps get recognized by name Data that reaches freddy through Apple Health or Health Connect arrives stamped with the name of whichever app wrote it. freddy recognized only a short list of those names, so plenty of real sources showed up as an unfamiliar package name or a bare string. That list is now much longer. Entries resolve to the actual app where the name is known, and sync or relay apps are labeled as such so it is clear the data passed through something before it landed. What gets ingested has not changed, only how it is attributed and displayed. Garmin history import handles Mac exports Uploading your Garmin export from a Mac could fail. That is fixed. A partial import now reports itself as partial rather than as a clean success, so you can tell when something did not make it through. Smaller fixes Polar rides recorded with a power meter now bring per-second power through, derived from the left and right crank sample streams. Concept2 backfills no longer stop at 250 results. Accounts with more history than that were being truncated after the first page. Connecting an AI client no longer fails when two authorization attempts arrive at the same moment. Coming next The Android app is due the same "still importing your history" status that Apple Health already shows, plus a prompt to exempt freddy from battery optimization so background syncing is not throttled by the system.

## What's new in freddy: week of Jul 13, 2026
URL: https://freddy.coach/articles/whats-new-2026-07-19
Garmin's full history import is now easy to find, two Apple Health data bugs are fixed, sync status stops claiming to be done early, and the MCP write tools were renamed.
Four changes landed in freddy this week: Garmin's full history import is now something you can actually find, two Apple Health bugs that were quietly distorting numbers are fixed, Apple Health's sync status stopped claiming to be done early, and the write tools your AI client calls got renamed. Garmin's full history import, now where you can see it Older Garmin history comes from the data export Garmin gives you on request. freddy has been able to import that export for a while, but it sat somewhere most people never looked. That changed this week. There are now three ways into it: A prompt right after you connect Garmin. An import card on the dashboard until you have done it. A direct link that drops you straight into the flow. Separately, Garmin backfills that were being blocked now trigger a permission refresh and retry, instead of stalling without saying anything. Apple Health numbers that add up Fixed two Apple Health number bugs. Sync status that tells the truth Apple Health used to flip from "Syncing" to done while it was still importing history. It now stays marked as syncing until the import genuinely finishes. The dashboard also explains why the import moves in chunks rather than all at once: Apple Health only lets the app read while your phone is open and unlocked, so history arrives in bursts whenever you pick your phone up. The write tools are now the source family The tools your AI client uses to manage connections are now connect_source, sync_source and disconnect_source. They used to be named after wearables, which never matched what freddy reads. Scales, CGMs, training platforms, gym logs and environmental sensors are all in there. Their descriptions were also written around one particular client, and are now client-agnostic, so they read sensibly in Claude, ChatGPT, Perplexity, or anything else that speaks MCP. If your client cached the old tool names, refresh the connection to pick up the new ones.

## What's new in freddy: week of Jul 6, 2026
URL: https://freddy.coach/articles/whats-new-2026-07-12
WHOOP is now generally available, ready-made prompts arrive in your AI client, and freddy can answer questions that span months or years in a single pass.
WHOOP is now generally available in freddy, and a handful of changes this week make your health data easier to ask about and quicker to arrive from your phone. WHOOP is now generally available WHOOP moved out of beta this week, which means anyone can connect it from the dashboard without requesting access first. Once connected, it brings your recovery, strain, sleep and HRV into freddy so your AI client can read them alongside everything else you track. Ready-made prompts in your AI client freddy now ships a set of ready-made prompts. In AI clients that surface them, such as Claude, you will see a weekly health review, a long-term trends view, and a guided walkthrough for connecting a new source. Instead of phrasing a detailed question from scratch, you can pick one of these and your AI runs the full query for you. It is a faster way in for anyone who is not sure how to ask. Questions that span months or years Long-range questions used to be awkward. The underlying data is recorded at a fine grain, so a two-year question would return far more rows than fit in a single answer. Your AI can now answer long-range questions, such as how your resting heart rate trended over two years. freddy also got clearer built-in guidance for paging through large date ranges, so long history questions come back complete rather than quietly cut short. Near-live data from your phone Data that arrives through Apple Health and Health Connect, including glucose and workouts, now shows up close to live. If one of those syncs ever stalls, it now recovers on its own and catches up without needing a manual nudge. Glucose now has daily summaries, including time in range. Smaller additions A few other things landed this week. Oura gains a daily breathing disturbance index, a sleep-quality signal derived from your overnight SpO2 readings. Health Connect can now read skin temperature deviation if your device records it. And the Garmin history import got a fix: sleep files that previously failed to parse now import correctly, so if you uploaded a Garmin data export and your sleep was missing, re-uploading will pick it up. Coming next A read-only API is in early access, so developers and power users can pull their own freddy data programmatically. It is opening up more widely soon.

## What's new in freddy: week of Jun 29, 2026
URL: https://freddy.coach/articles/whats-new-2026-07-05
freddy is now on Android with Health Connect, queries work across more AI clients, Suunto workout calories are fixed, and morning data lands on time.
freddy landed on Android this week, cleaned up how queries arrive from different AI clients, corrected Suunto workout calories, and got fresh data flowing on time in the morning again. freddy is now on Android The Android app is live on the Google Play Store, open to everyone. It reads Health Connect, so the health and training data already on your phone can flow into freddy and become something you ask your AI client about, whether that is Claude, ChatGPT, Perplexity, or another. Through Health Connect you can bring in Samsung Health, Google Fit, and Fitbit. Connecting Health Connect no longer requires an access request, it is open to every member. Health Connect workouts also come through with more detail now. Cadence and per-segment and per-lap data are read, so a run or ride keeps its structure instead of collapsing to a single summary. Queries work across more AI clients Some AI clients send tool arguments as plain text instead of structured values, and that was quietly breaking metric queries from those clients. freddy now accepts arguments in either shape, so asking for your metrics works no matter which client you use. Suunto workout calories are correct now Suunto workout calories showed the wrong numbers. Suunto workout calories now read correctly, so the values line up again. Fresh data lands on time in the morning Webhook sources like Polar and Oura could show up an hour or two late in the morning. That is fixed, so fresh data from those sources arrives on time. More Polar history Polar can now bring in more than three years of history. To pull the older history into a connection you already have, resync it from the dashboard. Attach images in support chat You can now attach an image when you message support in the app. A screenshot of what you are seeing often explains a problem faster than a paragraph.

## How Garmin, WHOOP, Oura, Polar, Ultrahuman, Withings, Amazfit, and Eight Sleep measure HRV
URL: https://freddy.coach/articles/why-wearables-report-different-hrv-numbers
One body, seven wearables, HRV numbers from 38 to 95. A deep look at RMSSD vs SDNN and how WHOOP, Oura, Garmin, Fitbit, Polar, Ultrahuman, Apple, and Withings each derive their figure.
One body, one recovery state, and a pile of heart-rate-variability numbers that refuse to agree. On matched nights held at the same recovery state, one person’s own devices reported HRV anywhere from 38 to 95. Apple Health said 95, Ultrahuman said 75, Polar said 52, WHOOP said 46, Garmin said 45, Oura said 40, and Fitbit said 38. Everything there is RMSSD except Apple, which is SDNN. The strangest part sits inside the list, not next to it. The same Ultrahuman ring showed 75 in its own app and wrote 95 into Apple Health for the same night. So the real question is not which device is right. It is why the numbers are so different, and the answer is device by device. First, why two numbers can describe the same heartbeats HRV is not one measurement. It is a family of ways to summarize the tiny timing gaps between your heartbeats. Consumer devices use two of them, and they are just two different formulas run on the same beats. RMSSD looks at how much each beat differs from the one right before it, so it catches the fast, moment-to-moment wobble. SDNN looks at the total spread of all those gaps across the whole recording, so it soaks up slower rhythms like your breathing and body clock too. Because they measure different things, SDNN usually reads higher than RMSSD from the same person, which is most of why Apple’s 95 sits at the top. The two were never meant to match, so lining them up side by side is comparing apples to a slightly different fruit. WHOOP WHOOP measures from the optical sensor on your wrist and reports RMSSD in milliseconds, which its developer API confirms as the hrv_rmssd_milli field. The twist is timing. WHOOP does not average the whole night. Its own education material and patent describe a sleep-weighted figure that leans on your last big slow-wave-sleep episode before you wake, the theory being that deep sleep is the cleanest window on recovery. So a WHOOP HRV is really “recovery HRV from late in the night,” not a flat overnight mean. Oura Oura reads from an infrared sensor against your finger and reports RMSSD. Rather than take one reading, it samples in five-minute chunks across the whole night and reports the average of them all. That whole-night average is the raw nightly number. The friendlier “HRV Balance” layer on top compares a rolling two-week average to your three-month baseline, using nighttime HRV only. Oura is one of the more transparent devices here, and it spells all of this out in its own support pages. Garmin Garmin measures optically at the wrist, only while you sleep, and reports HRV Status as a seven-day rolling average compared against a personal baseline that takes about three weeks of nights to settle. Garmin’s own docs are clear on the what and the when but never actually print the metric name. It is almost certainly RMSSD, because Garmin bought Firstbeat in 2020 and Firstbeat’s engine is well documented as RMSSD-based, but that last step is industry consensus rather than a line in Garmin’s manual. Polar Polar is the most upfront of the bunch. It names RMSSD explicitly, measures optically at the wrist over roughly the first four hours of sleep as part of Nightly Recharge, and compares the result to a baseline built from your previous 28 nights. Polar even explains why it uses the early hours rather than the whole night, arguing those hours reflect recovery more sensitively than a full-night average. So a Polar number and an Oura number can disagree simply because one watched the front of the night and the other watched all of it. Ultrahuman Ultrahuman reads from a finger sensor and, by its own account, works with both RMSSD and SDNN. In its own app it shows an RMSSD-style figure, which is the 75 above. When it hands data to Apple Health, though, that data lands in Apple’s single HRV slot, which is SDNN, so the same night resurfaces as 95. That is the whole “one ring, two numbers” mystery. It is not a glitch, it is two different metrics wearing the same label. Ultrahuman does not publish its exact windowing, and its own study notes HRV windowing was not studied, so the finer details here are reasonable inference rather than documented fact. Fitbit Fitbit reads optically at the wrist and states plainly that it uses RMSSD, measured during sleep, and needs a sleep block longer than three hours to report anything. The precise construction, the median of five-minute RMSSD windows taken from your longest qualifying sleep period, comes from a peer-reviewed validation of the Versa 4 rather than Fitbit’s own help pages. The practical point is that a Fitbit number is a median of chunks, not a single flat average, which nudges it away from what a whole-night-averaging device would report. Apple Watch and Apple Health Apple is the deliberate outlier. Apple Health stores HRV under exactly one type, SDNN, in milliseconds, so every HRV reading in the Health app is an SDNN value no matter which app wrote it. The Apple Watch also does not compute one overnight average. It takes short reads, on the order of a minute, when you are still, including background samples overnight and during Breathe or Mindfulness sessions, so a single day can hold several SDNN samples. Between SDNN reading higher than RMSSD and those short sampling windows, Apple numbers routinely sit above the RMSSD crowd, which is exactly what the 95 at the top shows. Withings Withings reports RMSSD and takes an unusual approach, reporting separate HRV for the first and last stretches of sleep rather than folding the night into one figure. Its ScanWatch line also carries an FDA-cleared single-lead ECG you trigger yourself for a clinical-grade 30-second reading, which is a different animal from the passive optical measurement it does overnight. The RMSSD choice and the split-window approach are documented in reputable secondary sources, though Withings does not publish its exact baseline math. Amazfit and Zepp Amazfit, whose app is Zepp, reports RMSSD measured overnight through its BioTracker optical sensor at the wrist, and compares it against a baseline drawn from roughly the last seven days of sleep HRV. This one is straightforward and documented on Amazfit’s own pages. It sits in the same RMSSD, wrist-optical, whole-night family as Garmin and Fitbit, which is why devices from that group tend to cluster closer together than any of them do to Apple. Eight Sleep Eight Sleep never touches your skin. Its Pod sits under you and reads HRV contact-free through the mattress, detecting the tiny movements your body makes with each heartbeat, a technique called ballistocardiography. It reports RMSSD in milliseconds and says it has validated that against ECG across hundreds of nights. So a bed and a finger ring can both hand you an RMSSD number, arrived at through completely different physics, which is a good reminder that the sensor and its placement shape the result as much as the formula does. What the number is actually worth Line those up and the 38-to-95 spread stops looking like a fault. Different metric, different slice of the night, different sensor, different smoothing, and even the six RMSSD devices ran from 38 to 75. None of them is wrong. Each is internally consistent. They are simply not measuring the same thing. The useful takeaway is that the absolute number means very little on its own. HRV is a signal only as your own trend, on one device, against your own baseline. Switching devices resets that baseline, which is why a “drop” the week you change rings is usually telling you about the ring, not about you. This is why freddy keeps each source in its own unit and its own baseline and never quietly blends them. When your AI reads your data, whether that is Claude, ChatGPT, or anything else, it sees “Oura RMSSD, whole-night average” and “Apple Health SDNN, periodic reads” as the distinct measurements they are, and compares your Oura trend to your Oura history and your Apple trend to your Apple history. That is the honest version of the question you actually care about, which was never “is 95 better than 46,” it was “is tonight better than my normal.” Method details verified 2026-07-02 against provider documentation and peer-reviewed validation studies. Check vendor sites for current details, and treat HRV as trend information, not medical advice. Sources RMSSD vs SDNN fundamentals Shaffer & Ginsberg 2017, An Overview of Heart Rate Variability Metrics and Norms: PMC full text Task Force of the ESC/NASPE 1996, Heart rate variability standards: European Heart Journal WHOOP WHOOP developer docs, recovery object (hrv_rmssd_milli): developer.whoop.com WHOOP patent, HRV timing and slow-wave / last sleep phase: USPTO Oura Oura Support, Heart Rate Variability (RMSSD, 5-min samples, whole-night average): support.ouraring.com Oura Blog, HRV Balance (2-week rolling avg vs 3-month baseline): ouraring.com Garmin Garmin blog, Understanding HRV Status (sleep-only, 7-day average, 3-week baseline): garmin.com Third-party detail on RMSSD via the Firstbeat engine: biologyinsights.com Polar Polar Support, Nightly Recharge recovery measurement (RMSSD, wrist, first ~4 hours, 28-day baseline): support.polar.com Polar Blog, autonomic nervous system (ANS Charge over the first four hours): polar.com Ultrahuman Ultrahuman Blog, RMSSD vs SDNN HRV Metrics Explained: ultrahuman.com Ultrahuman, accessing HRV in Apple Health: ultrahuman.com Fitbit Fitbit Help, HRV (uses the common formula RMSSD, during sleep, >3h): help.fitbit.com Peer-reviewed Versa 4 validation (median of 5-min RMSSD epochs): PMC full text Apple Watch and Apple Health Apple Developer, HealthKit heartRateVariabilitySDNN (SDNN is HealthKit’s single HRV type): developer.apple.com Terra Research, How HRV Actually Works (Apple SDNN vs RMSSD elsewhere): tryterra.co Withings Withings Support, ScanWatch 2 HRV: support.withings.com Terra Research (Withings reports separate HRV for the first and last stretches of sleep, RMSSD): tryterra.co Amazfit and Zepp Amazfit, What is Heart Rate Variability (RMSSD, overnight, ~7-day baseline): us.amazfit.com Amazfit health technology (BioTracker PPG optical sensor): us.amazfit.com Eight Sleep Eight Sleep, Pod Heart Rate and HRV Accuracy (RMSSD, contact-free, ECG-validated): eightsleep.com Ballistocardiography HRV validation (contact-free nocturnal HRV method): JMIR Biomedical Engineering

## What's new in freddy: week of June 22, 2026
URL: https://freddy.coach/articles/whats-new-2026-06-28
freddy is now an official ChatGPT app, Ultrahuman is generally available, intervals.icu workouts include per-lap detail, and the dashboard shows each source's sync status more clearly.
This week brought an official ChatGPT app, Ultrahuman out of beta, more detailed intervals.icu workouts, and a clearer dashboard. An official ChatGPT app freddy is now an official ChatGPT app. Connecting it to ChatGPT used to mean pasting a custom connector URL in by hand. It can now be added straight from the ChatGPT app directory, on mobile as well as desktop. Nothing changes for anyone using freddy with Claude, Perplexity, or another client. This just makes the ChatGPT path much shorter. Ultrahuman is now generally available Ultrahuman has moved out of beta, so anyone can connect it without requesting access first. freddy now parses the full Ultrahuman sleep record too, so sleep stages, timing, and the rest of the detail Ultrahuman captures come through to your AI alongside your other connected sources. Richer workout detail from intervals.icu intervals.icu workouts now come through with their individual intervals, or laps, not just the workout total. That makes it possible to look at how each rep, split, or effort went within a single session, instead of only the session summary. A clearer dashboard The connection tiles on the dashboard now show what each source is doing right now: actively syncing, up to date, or in need of attention. The sync button locks while a sync is already running, so it cannot be triggered twice by accident. New accounts also get a setup checklist on the overview that walks through connecting a source and linking an AI client. Recent data shows up first When freddy imports a long history from Polar, it now pulls the most recent training first, so your latest sessions appear while older history fills in behind them. There is no waiting on years of backfill before you can ask about last week. A fix for connecting Garmin Connecting Garmin partway through a day used to skip activities recorded earlier that day, because the first import stopped at midnight instead of the moment you connected. That window is now covered, so the day you connect comes in complete. Coming next Apple Health on iOS and Health Connect on Android are both in final testing. Once they are ready, your phone will be able to push the health data it already collects into freddy directly.

## freddy is now an official ChatGPT plugin
URL: https://freddy.coach/articles/use-your-health-data-in-chatgpt
freddy is now in the ChatGPT Plugin Directory. Connect it in a couple of clicks and ask ChatGPT about your sleep, training, and recovery using your own devices.
freddy is now an official plugin in the ChatGPT Plugin Directory. You can connect your sleep, training, recovery, and other health data to ChatGPT in a couple of clicks, then ask about it in any chat and get answers built on your real numbers instead of guesses. Until now, using freddy in ChatGPT meant turning on developer mode and pasting a connector URL by hand. That step is gone. freddy is a listed plugin now, so you find it, connect it, and you are done. It works in ChatGPT on the web and in the ChatGPT mobile app, and it is available to any logged-in ChatGPT account. How to connect freddy You will need a freddy account with at least one data source connected and synced. If you have not set that up yet, start at freddy.coach . Then: Open the freddy plugin page in ChatGPT's Plugin Directory ( Open freddy in ChatGPT ) and select Connect. Sign in with your freddy email and approve the read-your-health-data permission. In a chat, turn freddy on from the plus menu or type @freddy, then ask your first question. That is the whole setup. For more detail and troubleshooting, see the step-by-step guide . What you can ask Once freddy is connected, ChatGPT can read whatever you have linked, whether that is an Oura ring, a Garmin watch, a WHOOP band, a Polar device, a CGM, or a gym log. A few prompts to try: How did I sleep last night, and how does it compare to my average? Summarize my last five workouts and the heart rate zones I trained in. My resting heart rate has been climbing this week. What might be going on? ChatGPT can also connect a new source, sync, or disconnect on your behalf, so you can manage freddy without leaving the conversation. freddy is read-only on every account you connect, and you can revoke access at any time. One connection, every assistant freddy speaks the Model Context Protocol, so the same account works in Claude, Perplexity, your own agents, and now ChatGPT as a listed plugin. Connect your sources once and ask whichever assistant you already use. freddy is free to start at freddy.coach . Setup steps verified June 26, 2026.

## What's new in freddy: week of June 8, 2026
URL: https://freddy.coach/articles/whats-new-2026-06-14
Two new sources (Runalyze and Lyfta), full Garmin history import from your data export, pause and resume for any connection, and a connected-devices view.
Here is what shipped in freddy during the week of June 8: two new sources, a way to bring in your full Garmin history, and more control over your connections. Two new sources Runalyze (beta), the endurance and health analytics platform, bringing in your training load, fitness and fatigue trend, and sleep and HRV if you track them there. It needs a Runalyze Supporter or Premium account for API access. Lyfta (beta), a strength training tracker, so your lifts and workout history come through. Both connect with an API key and are in beta. You can request access from your dashboard. Import your full Garmin history You can now import your Garmin export for older history: upload the data export Garmin gives you, and freddy imports the full history from that file. This week added eleven more data families to the importer, including endurance and hill scores, blood pressure, hydration, and ECG. If you imported earlier, re-upload to pick up the new ones. Pause and resume any connection You can now pause a source instead of disconnecting it. Pausing stops syncing but keeps all of your history, so you can resume later without re-importing anything. It is handy when you are switching devices or want to mute one source for a while. Smarter reconnect The Reconnect button now only appears when a connection genuinely needs re-authorization, instead of showing when everything is fine. If a connection does drop, you also get a heads-up email so your data does not quietly go stale. See where you are signed in A new settings view lists every active login on your account and lets you revoke any of them. For people building with freddy freddy is now listed in the official MCP registry, so it shows up when AI clients browse for connectors. There is a new For Agents page covering the CLI and the headless setup. Apple Health is next. It is code complete and in final testing.

## What's new in freddy: week of June 1, 2026
URL: https://freddy.coach/articles/whats-new-2026-06-07
Garmin and Airthings reach general availability, deeper Polar data with heart-rate zones, richer Garmin activities, a redesigned dashboard, and the new freddy CLI.
This was one of the busiest weeks freddy has had. Two sources reached general availability, Polar and Garmin data got a lot deeper, the dashboard was rebuilt, and there is a new way to use freddy from a terminal. Garmin and Airthings are now generally available Garmin Connect is out of beta and open to everyone. Airthings is too, bringing indoor air quality and radon into freddy as the first environmental source, so your air can sit alongside your sleep and training. Deeper Polar data Polar History is now offered to every Polar user, which pulls in years of past training rather than just data from the moment you connected. freddy also now reads heart-rate time-in-zone, Z1 through Z5, for Polar workouts, so you can ask how much time you spent in each zone. Richer Garmin activities Garmin activities now include power, elevation, cadence, pace, and temperature, each attributed to the specific device the activity came from. SpO2, respiration, and stress are now captured too. A redesigned dashboard The dashboard was rebuilt around five clear areas: Overview, Connections, Data, Activity, and Settings, with support chat and a product updates view built in. freddy on the command line The new freddy CLI lets you sign in, connect a source, and query your data straight from a terminal. It is handy for scripts and for headless agents that do not use a chat window. Fixes Wahoo workouts now come through correctly.

## How to give an AI agent access to your health data from the command line
URL: https://freddy.coach/articles/ai-agent-health-data-command-line
A command-line tool lets a headless AI agent sign in, connect your wearables, and read your sleep, recovery, and training data, without you building an OAuth integration by hand.
Headless AI agents (Claude Code, Cursor, Cowork, your own bot) are good at multi-step work: pull some data, reason over it, take an action, check the result. Your wearable data is a natural thing to point one at. "Look at my last 30 days of sleep, find the worst nights, and tell me what they had in common" is exactly the kind of question an agent answers well. The friction was always the wiring. Giving an agent read access to your Oura or WHOOP or Polar data meant registering an OAuth client, walking a device-authorization flow, storing and refreshing tokens, and speaking the Model Context Protocol to actually fetch metrics. That is a real afternoon of work, and most of it is plumbing nobody wants to own. A command-line tool removes that afternoon. This piece walks through what it looks like to give an agent your health data in a couple of commands, what is happening underneath, and when you would still want to do it the long way instead. What "headless agent" means here A headless agent is one that runs without a chat UI you click around in: a coding agent in your terminal, a scheduled job, a script you wrote, a bot in a server somewhere. It can run shell commands but it cannot paste a URL into a "Connectors" settings field the way you would in a desktop chat app. That distinction matters, and it is why a CLI fits this case better than the paste-a-URL setup that consumer apps use. If you are using a desktop chat app instead (Claude, ChatGPT, Perplexity), you do not need any of this. You connect your wearable once on a dashboard, copy a personal URL, and paste it into the app. The command-line path below is for the headless case. The short version freddy is a hosted multi-wearable server that reads from your wearables' APIs and exposes them over MCP. Its CLI, published as @freddy-coach/cli , wraps the whole flow. From a terminal: npx @freddy-coach/cli login --scope "mcp account:read connections:write" npx @freddy-coach/cli connect oura npx @freddy-coach/cli query --metrics hrv,sleep_duration --days 7 The first command prints a sign-in URL; you open it in a browser, approve, and the CLI stores your tokens. The second prints a URL to authorize your wearable. The third reads your data. Swap oura for whoop , polar , withings , strava , and others. The CLI is free and open source under the MIT license. An agent runs the same commands. Because every command takes a --json flag that prints a single machine-readable object, the agent gets structured output it can reason over directly: freddy query --metrics hrv,sleep_duration,resting_heart_rate --days 30 --json The agent never sees a raw token or touches an OAuth redirect. It just runs commands. What is actually happening underneath Nothing magic, just the standard pieces wired together for you: Device-flow sign-in. The CLI uses the OAuth 2.0 Device Authorization Grant (RFC 8628), the same flow a smart TV or a developer CLI uses when there is no reliable browser callback. It registers a client, requests a code, shows you a URL, and polls until you approve. No redirect_uri to configure, which is exactly why it suits headless tools. Token storage and refresh. The CLI signs in once and keeps you signed in. MCP for the data. Queries go through the Model Context Protocol, the open standard AI clients use to pull data from external sources. The CLI calls the same query_metrics and list_metrics tools an MCP-speaking chat app would, and hands you the result. Run the metrics command to see exactly which metrics your connected wearables expose, with their date ranges and units, before you query. Scopes, and what the agent cannot do Access is scoped. The default sign-in requests read access ( mcp account:read ), which is enough to query your data and read your profile. Connecting or disconnecting wearables needs a broader scope ( connections:write ) that you request explicitly at sign-in. This means you can hand an agent a read-only session and know it cannot change your connections. Deleting your account is only possible from the dashboard. You can review and revoke any agent's access from your dashboard at any time, which immediately invalidates its tokens. When you would still do it the long way A hosted CLI is the fast path, not the only one. Two honest alternatives: Wire the flow yourself. If you are building a product and want to embed the sign-in rather than shell out to a tool, the device flow, the REST surface, and the MCP tools are all documented. You do the same work the CLI does, with full control over the UX. This is the right call if the integration is part of something you ship. Self-host the server. If you would rather not have a third party hold a copy of your data, Open Wearables is a community project that runs the wearable-to-MCP bridge on hardware you control. You handle the OAuth setup and keep the server running. This is the right call if you self-host other things and want your health data to stay on your own machine. freddy is hosted, so there is nothing to run yourself. Pick based on which tradeoff you prefer. For most people who just want their agent to read their wearable data today, the CLI is the least-effort path that works. Cost The CLI itself is free to install and free to use. freddy is free for a single wearable and $49/year for unlimited connections and full history on the paid tier. The AI agent on the other end has its own pricing, separate from any of this. Where to start If you already have a wearable, install the CLI, run login with the connection scope, connect your device, and query . Five minutes, no integration code. From there, point your agent at the same commands and let it pull what it needs. If you are choosing a wearable with good developer support, the comparison of wearable APIs for AI agents covers which devices expose the richest data. And if you want the broader picture of how wearables, AI clients, and MCP servers fit together, start with connecting wearable data to AI . Updated 2026-06-01. Pricing and feature details verified 2026-06-01. Check vendor sites for current details before committing.

## What's new in freddy: week of May 25, 2026
URL: https://freddy.coach/articles/whats-new-2026-05-31
Ultrahuman joins in beta, Wahoo and Concept2 reach general availability, a refreshed brand, and connections that stop dropping on a brief hiccup.
Here is what changed in freddy during the week of May 25: a new source, two graduations out of beta, a fresh look, and a reliability fix you will feel. New and graduated sources Ultrahuman joins in private beta, covering the Ring AIR and the M1 glucose sensor. Wahoo and Concept2 are now generally available, so you can connect them without requesting access first. A refreshed look freddy has a new brand mark. You will see the updated icon across the app and in emails. Connections that stay connected freddy now confirms a connection is really dead before disconnecting it. A brief provider outage or a single failed request no longer drops your source and forces you to reconnect, which is how a few people lost connections in the past.

## What's new in freddy: week of May 18, 2026
URL: https://freddy.coach/articles/whats-new-2026-05-24
Multi-device tracking that no longer loses data, more reliable Garmin history, and cleaner account disconnects.
This was a quieter, under-the-hood week in freddy, focused on getting multi-device tracking right and making history load more reliably. Track more than one device without losing data If you wear more than one device, say a Garmin and an Oura, freddy now tags each metric with the device it came from. Overlapping readings no longer overwrite each other, so a day with two sources keeps both. The dashboard data viewer now shows a device column too, so you can see exactly where each number originated. More of your Garmin history, more reliably Garmin history now loads recent days first and then works backward through your past in chunks. You see current data quickly, and a large history loads steadily in the background instead of stalling. Cleaner disconnects When you delete your account, freddy now revokes its access to your connected wearables before it wipes your data, so nothing is left authorized after you leave.

## Polar versus Oura for everyday recovery tracking
URL: https://freddy.coach/articles/polar-versus-oura-for-everyday-recovery-tracking
A head-to-head comparison of Polar Vantage V3 and Oura Ring Gen 4 for recovery tracking: form factor, cost model, Nightly Recharge vs Readiness Score, and which to buy.
Polar and Oura are the two devices most people choosing their first serious recovery wearable end up comparing. They aim at different people. Polar builds outward from sport: a wrist computer for athletes who log workouts and also want overnight recovery insight. Oura builds inward from sleep: a ring that focuses entirely on resting-state signals, with no workout controls and no screen. The right answer depends less on which data is better and more on which starting point fits the way you actually live. Polar Vantage V3 vs Oura Ring Gen 4 at a glance Polar Vantage V3 Oura Ring Gen 4 Form factor Wrist watch Ring Starting price $699.99 $349 Ongoing cost None $69.99/year subscription Battery 10 days 8 days Recovery signal Nightly Recharge (-10 to +10) Readiness Score (0-100) HRV measurement RMSSD RMSSD What you're putting on your body Polar Vantage V3 is a full-size sport watch with a color display, onboard GPS, and physical button controls. It is a device you will notice on your wrist. That is an advantage if you also use it during workouts: you can see pace, heart rate, and navigation in real time. During sleep, the optical sensor on the underside measures the overnight signals that feed Nightly Recharge. Oura Ring Gen 4 is a lightweight titanium ring you wear on a finger. It has no screen, no buttons, and no GPS. That passivity is the whole proposition: it reads your body without asking anything of you. The tradeoff is that it is not a workout device. You can log a workout in the Oura app, but there is no live heart rate display and no sport profiles. If you also do heavy strength training, move the ring to a less prominent finger to avoid pressure discomfort during barbell grips. What it costs Polar Vantage V3 is $699.99 upfront, with no ongoing subscription. The data is yours, processed through the Polar Flow app. Polar's lineup also includes lower-priced models if the Vantage V3 is more than you need. Oura Ring Gen 4 starts at $349 for the hardware, plus a $69.99/year membership for the full app experience, including the Readiness Score, sleep staging, and HRV trends. The membership is required for the features that make the ring worth buying. The gap in headline price narrows when you account for the subscription accumulating over time. The recovery signal Polar's recovery signal is called Nightly Recharge, and it has two components. ANS charge measures how well your autonomic nervous system calmed down during roughly the first four hours of sleep, using heart rate, HRV (RMSSD), and breathing rate. Sleep charge reflects total sleep duration and cycle quality. Both are expressed relative to your personal 28-day baseline and combined into a score from -10 to +10, where zero is your typical level. A strong positive means your system recovered more than usual; a negative means it did not. Oura's Readiness Score is a 0-100 composite built from seven contributors: resting heart rate, HRV balance (comparing your two-week trend to your long-term average), body temperature deviation from your personal baseline, previous day activity, activity balance over 14 days, last night's sleep, and sleep balance over time. A score of 85 or higher is read as ready for a demanding day. Because it incorporates body temperature and multi-week activity patterns, Readiness reacts to things Nightly Recharge does not: a sustained temperature deviation can signal early illness, and the activity balance contributors penalize overtraining before your performance declines. Neither signal is more accurate in an absolute sense; they model recovery differently. Nightly Recharge is scoped to what happened during last night's sleep, which makes the result direct to interpret. Readiness is broader, making it sensitive to multi-day patterns but harder to trace back to a single cause. Prefer Nightly Recharge if you want a signal tied to each individual night. Prefer Readiness if you want a composite view that includes what you did yesterday and how your month is trending. Sleep tracking Both devices track sleep stages (light, deep, and REM) using optical sensors and accelerometers. Oura additionally monitors skin temperature throughout the night. A deviation above your personal baseline surfaces in the Readiness Score and as a standalone trend in the app, and can serve as an early indicator that something is off before other symptoms appear. Polar incorporates breathing rate as a separate input into the ANS charge calculation, a data point Oura does not surface separately in sleep reporting. Both Polar and Oura measure HRV as RMSSD, so you can compare trends across your own history on either device. Cross-device comparison is not meaningful: different sensor positions and sampling algorithms produce different absolute values for the same person at the same time. Connecting to an AI assistant Both Polar and Oura are supported by Freddy ($49/year), a personal MCP server that connects your wearable data to any MCP-capable AI client, including Claude, ChatGPT, and Cursor. Once connected, you can ask questions across your full history rather than reading individual days in the device app. Which one to buy Buy Polar if you train seriously and want one device to cover both workouts and recovery. The GPS, sport profiles, and training load features are genuinely useful for athletes, and the no-subscription pricing is a real advantage if you want to avoid ongoing fees. Buy Oura if your focus is sleep and daily readiness. The ring form factor is comfortable overnight, disappears during the day, and stacks cleanly with a smartwatch if you wear one. The body temperature signal and the breadth of the Readiness Score make it the more comprehensive passive tracker of the two. Do not buy either as your only device if real-time GPS feedback during outdoor exercise is a priority: Oura has none, and buying a Vantage V3 purely for sleep tracking means paying for sport features you will not use. Pricing and feature details verified 2026-05-19. Check vendor sites for current details before committing.

## What's new in freddy: week of May 11, 2026
URL: https://freddy.coach/articles/whats-new-2026-05-17
Three new data sources (Garmin, Wahoo, Concept2), a simpler email-code sign in, full-history queries, and cleaner Perplexity support.
Here is what shipped in freddy during the week of May 11. It was a big week for new sources: three of them, plus a simpler way to sign in and a handful of fixes that make your history more complete. Three new data sources freddy can now connect to three more sources, all in private beta: Garmin Connect, the most-requested source, covering activities, sleep, and recovery. Wahoo, for rides and workouts. Concept2 Logbook, for rowing, SkiErg, and BikeErg. All three are in beta. You can request access from your dashboard. A simpler way to sign in freddy now signs you in with a six-digit code sent to your email, instead of a magic link. Typing a code back into the same browser avoids the problem magic links had on phones, where the link opened in the email app's browser rather than the one you started in. Ask about any stretch of your history The query tool now takes explicit start and end dates, so your AI client (Claude, ChatGPT, Perplexity, and others) can ask about any window of your data, not just the most recent days. Fixes Connecting freddy through Perplexity now works without the sign-up hiccups some people hit. Withings sleep now includes stage durations, and body temperature comes through. More next week.

## Best wearable APIs for AI agents: a developer's guide
URL: https://freddy.coach/articles/best-wearable-apis-for-ai-agents
Practical comparison of the major wearable APIs for developers building AI tooling: data depth, OAuth quirks, and which APIs make AI integration painless versus painful.
If you're building anything that connects health data to AI, your first practical question is which wearable APIs are actually pleasant to integrate against. This piece is from someone who's read all of them. The criteria that matter: Documented and stable. Public docs, predictable shapes, no quiet breaking changes. OAuth not weird. Standard authorization code flow with refresh tokens, sane scopes. Useful data. Sleep stages, HRV, workouts, daily activity at minimum. Webhook support. So you can react to new data without polling. Here's how the major ones stack up. 1. WHOOP Developer portal . HRV in ms_rmssd. WHOOP's API is the cleanest of the big five. OAuth 2.0 flow is standard, scopes are sensible, the data shapes are well-documented, and you get sleep, recovery, strain, workouts, and HRV in a consistent JSON schema. Webhooks notify you when new data lands. The downside: subscription-only ($239/year), so anyone you're integrating for has to be a paying WHOOP user. The data depth is excellent if they are. 2. Polar AccessLink API . HRV in ms_rmssd. Polar's AccessLink is older and less polished than WHOOP's developer portal, but the underlying API is solid. Sleep stages, training sessions, daily activity, Nightly Recharge, Cardio Load, SleepWise data; depending on the model, you get a lot. Vantage V3 and Pacer Pro expose the full set; entry trackers expose less. Webhook support is real but the documentation is lean. OAuth is standard. 3. Oura Cloud API . HRV in ms_rmssd. Oura's API is well-documented and the v2 endpoints are stable. You get sleep, readiness, activity, heart rate, HRV, and workouts. OAuth is standard. The catch is the subscription tier. The hardware is $349, but the API returns data only for users with an Oura membership ($69.99/year). Without it, Gen3 and later rings return no data through the API. 4. Withings Developer API . HRV in ms_rmssd. Withings' API covers a wider range than the others because their hardware does too: scales, sleep mats, blood pressure cuffs, watches. The developer experience is competent, OAuth standard, data shapes consistent. The minor friction is that Withings auth flows have historically had region-specific quirks (EU vs US OAuth domains). Worth testing in production once before launching anywhere. 5. Suunto Developer portal . HRV in ms_rmssd. Suunto's API is multisport-focused: workouts, training metrics, occasional sleep. Less depth than the top four for general health, but if you're building something endurance-specific it's the right fit. OAuth standard, docs reasonable. Honorable mentions Hevy (https://api.hevyapp.com/docs): not a wearable but a strength-training app with a public API. Useful if you're building anything that integrates lifts. Intervals.icu (https://intervals.icu/api/v1/docs): aggregator that pulls from Garmin, Strava, etc. Useful if you want endurance analysis as a single feed. Dexcom (https://developer.dexcom.com/): CGM, not a wearable in the traditional sense, but their API is the standard for glucose data. Apple HealthKit : not a web API, iOS-only. If you're building inside an iOS app it's the right path. If you're building a web service, you can't use it directly. What about MCP-shaped APIs? If you're specifically building for AI agents, the MCP-shaped path is to put a server in front of these APIs and expose a single uniform tool surface. That's what Freddy does. It reads from Polar, Oura, WHOOP, Withings, Suunto, Hevy, Intervals.icu, and Dexcom and exposes them to any MCP client. Saves you from writing six OAuth flows yourself. If you'd rather build the integration yourself, all of the APIs above are fine starting points. WHOOP and Oura are the easiest to get running. APIs evolve; check the linked developer portals before committing implementation work.

## The complete guide to connecting wearable data to AI
URL: https://freddy.coach/articles/connecting-wearable-data-to-ai
What it takes to ask AI assistants like Claude or ChatGPT about your sleep, recovery, workouts, and HRV from any wearable. The pieces that have to fit together, and where each piece comes from.
You wear a Polar, Oura, WHOOP, or Garmin. You also use Claude, ChatGPT, or Perplexity. You've probably wondered why these two pieces of your life don't talk to each other. The short answer: they can. The path is just slightly indirect, and once you understand what's happening, the rest is mechanical. The three pieces To ask AI a question about your wearable data, three things need to fit together: 1. The wearable. It records the data and exposes it through a developer API. Polar has AccessLink, WHOOP has its developer portal, Oura has its Cloud API, and so on. The data is yours; the wearable just stores and serves it. 2. The AI client. Claude Desktop, ChatGPT, Perplexity, and others all support a protocol called MCP (Model Context Protocol). MCP is the open standard that lets AI assistants pull data from external sources during a conversation. 3. An MCP server in the middle. This is the bridge. It reads from the wearable's API on one side and exposes the data to MCP-speaking AI clients on the other. Without it, the AI client has no way to read your wearable; with it, asking the AI about your sleep is the same as asking it about a document you've shared. That's all there is. AI on one side, wearable on the other, MCP server in the middle. What an MCP server actually does An MCP server is a small program that exposes "tools" to AI clients. Tools are typed function calls the AI can make: list_metrics , query_metrics , get_workout_summary . When you ask the AI a question that needs your data, it calls the appropriate tool and gets back structured information it can reason about. A health-focused MCP server connects to multiple wearable APIs (because most people own more than one wearable, or move between them over time), normalizes the data into one consistent surface, and serves it through MCP. The AI doesn't need to know which wearable produced which data; it just calls the tools and answers your question. Why this is suddenly possible Two things changed in 2024-2025 that made this work: MCP shipped. Anthropic published the Model Context Protocol in late 2024. By 2025, ChatGPT, Perplexity, and several other AI assistants had implemented it. The protocol is open, so anyone can build a client or a server. Wearable APIs matured. The major wearable companies (Polar, Oura, WHOOP, Garmin, Withings, Suunto) all have OAuth-based APIs that you can authorize once and pull data from continuously. None of this required hacking; it just required someone to build the bridge. Three ways to set this up Option 1: Use a hosted MCP server The fastest path. A service like Freddy handles all of the OAuth flows, data sync, and MCP serving for you. You sign up, connect your wearables, and paste a personal URL into your AI client. Five minutes, no code. Cost: typically a small subscription. Freddy is $49/year on the paid tier and free for one wearable. This is the right choice if you want to start using your data with AI today. Option 2: Self-host an MCP server If you'd rather run the server yourself, Open Wearables is a community project that does roughly the same job as a hosted service, on hardware you control. You handle OAuth setup, you store the data, you keep the server running. Cost: free, plus the time to maintain it. This is the right choice if you self-host other things in your life and want to add this to the stack. Option 3: Build it yourself Every wearable's API is documented and standard OAuth. The MCP spec is also documented. If you're a developer who wants to wire one or two specific data sources into a specific AI assistant, you can write the integration in a weekend. Cost: developer time. Useful if you're building something specific (an iOS app that shows your wearable data via Apple's on-device AI, say) rather than a general-purpose tool. What conversations look like once it's set up Once you've chosen a path and connected your data, asking the AI questions is exactly like asking it about anything else. Some examples that work well: "How was my sleep last week?" "Compare my HRV this month to last month." "On nights I trained late, did my sleep score drop?" "What's my average resting heart rate trend over the past 90 days?" The AI pulls from the MCP server, reads the relevant metrics, and answers in natural language. You can ask follow-ups. You can ask it to look for correlations. You can paste an article you read about sleep and ask whether your data agrees. Common questions Is my data private? Depends on the path. Hosted services hold a copy of your data on their servers (with whatever security their privacy policy describes). Self-hosted means it stays on your machine. Building it yourself means you decide. None of the wearable APIs share your data with the AI provider directly; the data flows through whatever MCP server you're running. Can I use this with multiple wearables at once? Yes. Multi-wearable MCP servers (Freddy is one) read from all your authorized devices and expose the combined surface. The AI can correlate across them. Does the AI keep a copy of my data? The AI reads your data during a conversation but doesn't typically retain it after. Each new conversation starts fresh and pulls again. Refer to your specific AI client's privacy documentation. Does this require a subscription to the AI? It depends on the client. In Claude, freddy is in the connector directory. In ChatGPT, freddy is an official plugin in the Plugin Directory. Perplexity has MCP on Pro. The MCP server itself can be free; the AI client typically has its own pricing. Where to start If you have a wearable already, the fastest first step is to sign up for a hosted MCP server, connect your wearable, and add it to your AI app. freddy is in Claude's connector directory and ChatGPT's Plugin Directory , so you add it from inside the app. From there, you can ask any question that involves your data and the AI will pull what it needs to answer. If you don't have a wearable yet and you're choosing one, the comparison of wearable APIs for AI agents covers which devices have the best developer support. For sleep specifically, the Oura vs WHOOP comparison covers the two most-AI-friendly devices in detail. Updated 2026-05-09.

## HRV across wearables: how to actually compare what your devices report
URL: https://freddy.coach/articles/hrv-across-wearables
Different wearables report HRV differently. Here's why your Oura number doesn't match your WHOOP number, what units to expect from each device, and how to compare across them anyway.
If you've worn two wearables on the same day and noticed they reported wildly different HRV values, you weren't imagining it. The signal is real but the number isn't standardized. Here's what each device is actually measuring and how to read across them. Quick answer There are two different HRV metrics in common use: RMSSD (root mean square of successive differences): the most common HRV unit, reported in milliseconds. Higher is generally better. SDNN (standard deviation of NN intervals): measures total variability, also in milliseconds. Reports a different number for the same heart, usually larger. Most modern wearables report RMSSD. Apple Watch reports SDNN. The two are different measures, so an Oura reading and an Apple Watch reading should not be set side by side. For the detail, see how each wearable measures HRV . What each major wearable reports Wearable HRV unit Oura ms_rmssd WHOOP ms_rmssd Polar ms_rmssd Garmin ms_rmssd Withings ms_rmssd Apple Watch ms_sdnn ( ms_rmssd and ms_sdnn are how Freddy's data file labels them; the numeric values are in milliseconds in both cases.) Why the same heart produces different numbers Three reasons: Sampling window. Oura averages overnight (six to eight hours of data, taken when you're asleep and motion is minimal). WHOOP samples during sleep too but uses a slightly different aggregation window. Polar's Vantage line samples nightly. Apple Watch samples opportunistically across the day. The signal you get is shaped by when and how it's pulled from the underlying inter-beat intervals. Algorithm. RMSSD weights short-term variability heavily; SDNN weights overall variability. Same beats, different math. RMSSD numbers are typically smaller than SDNN numbers from the same data. Sensor placement and noise. A finger ring has cleaner photoplethysmograph signal than a wrist watch. A chest strap (Polar H10) has cleaner electrocardiogram signal than either. The cleaner the signal, the more accurate the inter-beat-interval calculation, and the more stable the HRV reading. So how do you compare across devices? The honest answer is: don't, on a single-day basis. Cross-device comparison is unreliable on any given measurement. What works: Compare each device against itself over time. A 14-day rolling average of your Oura HRV is a meaningful baseline. So is a 14-day average from your WHOOP. Each is internally consistent. Compare directionally. If both devices show HRV trending down for a week, that's a real signal regardless of unit mismatch. Don't try to derive a "true HRV." No wearable gives you that. Polysomnography rooms in sleep labs do, with chest electrodes; consumer wearables approximate. What this means if you connect HRV data to AI When you ask an AI assistant like Claude or ChatGPT about your HRV across multiple wearables, the assistant should: Track each device's HRV trend separately, not pool them. Caveat the unit difference if you ask "is my Oura HRV high or low?" The answer depends on what's normal for you on Oura. Notice agreement: if Oura and WHOOP both show a downtrend, that's the most reliable signal you'll get. Freddy reads HRV from each wearable in its native unit and exposes them to AI clients along with the unit field, so the AI knows what it's reading. If you've ever had an AI confidently tell you your HRV is "low" without specifying which device or unit, that's the kind of error this prevents. Practical recommendation If HRV is the metric you care most about: For sleep-context HRV: ring or chest strap. Oura (ms_rmssd) or Polar with H10 (ms_rmssd). For 24-hour HRV monitoring: WHOOP (ms_rmssd) or Garmin (ms_rmssd). For research-grade work: chest strap and a separate analysis pipeline. Don't rely on consumer-device-derived HRV for anything clinical. Verified 2026-05-09. HRV measurement on consumer wearables is an active area; what's reported here will likely shift as algorithms update.

## Oura vs WHOOP for sleep tracking: which one should you actually buy?
URL: https://freddy.coach/articles/oura-vs-whoop-sleep-tracking
Honest comparison of Oura and WHOOP for sleep tracking. Form factor, battery, accuracy, what each one's API exposes, and which fits which kind of user.
If sleep tracking is the main reason you wear a wearable, Oura and WHOOP are the two devices most people end up choosing between. They both do sleep stages, both report HRV, both have respectable APIs that AI assistants can read. But they aim at different people, and the right answer depends on which you are. This is the comparison from the angle that matters most: what each one actually measures, what their APIs expose, and what it feels like to wear them every night. At a glance Oura WHOOP Form factor Ring (Gen 4) Wrist strap Hardware price $349 None (subscription includes hardware) Subscription $69.99/year $239/year Battery 8 days 14 days Sleep stages Yes Yes HRV unit ms_rmssd ms_rmssd Official API Cloud API Developer API Sleep stages: how each one works Both devices report time in light, deep, and REM sleep, plus awake time. They use different sensor configurations to do it. Oura uses an infrared photoplethysmograph and an accelerometer in the ring; WHOOP uses optical sensors plus a 3-axis accelerometer on the wrist. Neither matches polysomnography (the medical gold standard), but both correlate well enough for trend tracking. The practical difference is that Oura is reading from your finger, where signal-to-noise on heart rate is high and motion is low. WHOOP is reading from your wrist, where signal-to-noise is lower and your movements interfere more (typing, side sleeping, dragging your wrist across the bed). For sleep specifically, ring data tends to be cleaner. For training and effort, wrist data with continuous monitoring is fine. Battery and what it means in practice Oura goes about a week between charges. WHOOP 5.0 goes about two weeks. WHOOP charges with a battery pack that slides onto the strap, so you can charge while you wear it; Oura you take off, sit on a charger for an hour, put back on. Different ergonomics for the same problem. What the APIs expose Both have first-party developer APIs, both authorize via OAuth, both return sleep, HRV, recovery, and workout data through documented JSON endpoints. WHOOP exposes its proprietary "Recovery", "Strain", and "Sleep Performance" scores; Oura exposes "Sleep Score", "Readiness", and "Activity". The proprietary scores are interesting marketing, but if you connect either device to an AI assistant via an MCP server, you'll usually find the AI can construct equivalent answers from the underlying signals (resting HR, HRV trend, sleep efficiency) regardless of which scoring system the device uses. Freddy reads both APIs and makes the data available to Claude, ChatGPT, Perplexity, and any other MCP-speaking AI client. You can ask the same question of both data streams. Who should buy which Buy Oura if: You want the cleanest sleep signal you can get without a chest strap. You hate wrist wearables (most people who do). You're willing to pay $349 upfront and $69.99/year for the app features. Buy WHOOP if: You also care about training load and 24-hour effort tracking. You don't mind a wrist strap. You prefer a subscription-only price ($239/year, hardware included). Don't buy either if: You only want sleep data and you already wear an Apple Watch you trust. (Apple's HealthKit is iOS-only and not directly queryable by web AIs, but Series 10 sleep tracking is fine for trend purposes.) You want medical-grade sleep data. Both are wellness devices, not diagnostic tools. Connecting either to AI If you have either device and you'd like to ask Claude, ChatGPT, or Perplexity about your sleep, the easiest path is a hosted MCP server like Freddy which reads from both APIs and exposes them to any MCP-speaking client. Pick a wearable, sign up, paste a URL. Pricing and feature details verified 2026-05-09. Things move fast in this space; check vendor sites for current details before committing.

## Connect Garmin to Claude
URL: https://freddy.coach/help/garmin-claude
+ You can use your Garmin data in Claude by connecting your Garmin account to freddy, then adding freddy to Claude from Claude's connector directory. Then ask Claude about your own sleep, stress, Body Battery, HRV and workouts, and it answers from your real records. Garmin Connect holds what your Garmin watch or tracker records. freddy makes that data available to your own Claude, and it stays yours. TL;DR In freddy, open Sources, find Garmin Connect and select Connect. This works on the web and in the app. Sign in to Garmin, review the data categories, and switch on historical data if you want the history import. Back in freddy, let the first import run and check which dates have arrived. Open freddy in Claude's connector directory and select Connect to Claude. Sign in to freddy, approve, and ask Claude about your Garmin data. Free works for this: 1 source, and your AI's queries cover the last 7 days. No card. What you need A freddy account. Sign up free at freddy.coach . A Garmin Connect account that receives your Garmin device data. Sync your device to Garmin Connect first. Claude on the web (claude.ai), in the desktop app, or on your phone. You do not need a paid Claude plan to add freddy. On Claude Team and Enterprise, an Owner adds freddy for your organization first. Set up Garmin in Claude Part 1: Get Garmin into freddy Sync your Garmin device to Garmin Connect. freddy can only read what has reached Garmin Connect. In freddy, open Sources, find Garmin Connect and select Connect. This works on the web and in the app. Sign in to Garmin with the same account that receives your device data. In the app, a "Switch on Historical Data" sheet appears first. History stays off unless that switch is on, so turn it on if you want history, then select Continue to Garmin. Review the data categories and allow the ones you want to share. Back in freddy, let the first import progress and check the dates that have arrived on the Garmin source card. Want more than the recent window? Connecting pulls about 30 days. For older history, use Import account history or Import older Garmin data with your Garmin data export. See Import Garmin account history and FIT files . Part 2: Add freddy to Claude Open freddy in Claude's connector directory . You can also go to Customize, then Connectors, and search for freddy. Select Connect to Claude. Claude opens a freddy sign-in page. Sign in with the email you use for freddy, then approve. Back in Claude, start a new conversation (Part 3). Prefer your phone? Open the directory link in your mobile browser and connect there. Once connected, freddy also works in the Claude mobile app. Part 3: Ask your first question Start a new Claude conversation. If Claude does not use freddy, select + in the message box, then Connectors, and turn freddy on. Ask: "What Garmin data do you have for me, and how far back does each type go?" Claude can then show you the metrics and date ranges freddy holds, so you know what to ask next. 6 questions to ask Claude about your Garmin data Copy these as they are, or change the dates. Sleep: "How did I sleep last night on my Garmin? Give me duration and stages, and name the source." Stress: "Show my Garmin stress for the last 7 days: daily average, max and qualifier, and how long I spent in each stress band. Which days stand out, and what did my sleep and training look like around them?" Body Battery: "How much did my Garmin Body Battery charge and drain each day this week? Which days drained the most, and what came before them?" HRV: "Show my Garmin overnight HRV for the last 7 days: last night average, 5 minute high and status. Which nights stand out? Name the source and dates." Workouts: "List my last five Garmin workouts with duration, average and max heart rate, and distance, pace and elevation where recorded. Which one was the hardest?" Putting it together: "Line up my Garmin workouts with my sleep and Body Battery charge and drain this week. Do hard days show up the next morning?" Tip: ask Claude to name the source and date range it used. It makes the answer easy to check. On Free, ask about the last 7 days. The full imported history unlocks with Pro. If data is missing Start in Garmin Connect. Check that your device has uploaded the record there. freddy can only read what has reached Garmin Connect. Recent history is absent: check the historical data permission you gave when you connected, and the import status on the Garmin source card. Do not delete the source just to retry a running import. Months or years are missing: connecting pulls about 30 days, and syncing again cannot go further. Use Import account history or Import older Garmin data for older history. New records are missing: use Sync on the Garmin source in freddy after confirming they are in Garmin Connect. Claude says it has no data: check that the dates you asked about are inside your plan's history window, and that Claude is connected to the same freddy account. On Free, your AI's queries cover the last 7 days, and the full imported history unlocks with Pro. More in Understand sync and find missing data and Garmin Connect: connect, sync and troubleshoot . Your data, your Claude freddy makes your own Garmin data available to your own AI. You own it. You can pause a source, and if you delete your freddy account, your data is deleted. See Delete your account . Use Paused on the Garmin source to stop future sync and keep your stored history. Your records in Garmin Connect stay where they are. FAQ Can Claude read my Garmin data? Yes, through freddy. You connect your Garmin account to freddy, and Claude reads that data from freddy once you add freddy from Claude's connector directory. Claude does not connect to Garmin directly. Which Garmin account do I use? The one that receives your device data. You sign in to Garmin when you connect and choose which data categories to share. What Garmin data can I ask about? Daily activity, heart rate, sleep, stress and Body Battery where they are recorded and shared. HRV arrives as overnight RMSSD (last night average, 5 minute high and status). Stress comes as a daily average, max, qualifier and time in each stress band. Body Battery comes as daily charged and drained values. Workouts arrive with duration, average and max heart rate, plus distance, calories, pace and elevation where your device recorded them. freddy's sources page also lists steps. What you see depends on what your Garmin device records. How far back does Garmin data go? Connecting Garmin pulls about 30 days. On Free, your AI's queries cover the last 7 days, and the full imported history unlocks with Pro. For older history, use Import account history or Import older Garmin data with your Garmin export. See Import Garmin account history and FIT files . What does it cost? Free is $0: 1 source, and your AI's queries cover the last 7 days, no card. Pro is $49 a year, and Believer is $249 once for life. Paid plans have no source limit and unlock your full imported history. You do not need a paid Claude plan. See pricing . Who owns the data, and can I delete it? You do. freddy makes your own data available to your own AI. If you delete your freddy account, your data is deleted. Steps are in Delete your account . Related Garmin Connect: connect, sync and troubleshoot Import Garmin account history and FIT files Connect freddy to Claude Choose and connect a data source Understand sync and find missing data HRV across wearables: how to actually compare what your devices report Recipe: Why am I tired today? Sources freddy connects Pricing Using ChatGPT instead? Connect Garmin to ChatGPT freddy is an independent service and is not affiliated with, endorsed by, or sponsored by Garmin or Claude.

## Connect Garmin to ChatGPT
URL: https://freddy.coach/help/garmin-chatgpt
+ You can use your Garmin data in ChatGPT by connecting your Garmin account to freddy, then adding the freddy app in ChatGPT. Then ask ChatGPT about your own sleep, stress, Body Battery, HRV and workouts, and it answers from your real records. Garmin Connect holds what your Garmin watch or tracker records. freddy makes that data available to your own ChatGPT, and it stays yours. There is no custom connector to build and no URL to paste. TL;DR In freddy, open Sources, find Garmin Connect and select Connect. This works on the web and in the app. Sign in to Garmin, review the data categories, and switch on historical data if you want the history import. Back in freddy, let the first import run and check which dates have arrived. Open the freddy app in ChatGPT and select Connect. Sign in to freddy, approve, then type @freddy in any chat and ask about your Garmin data. Free works for this: 1 source, and your AI's queries cover the last 7 days. No card. What you need A freddy account. Sign up free at freddy.coach . A Garmin Connect account that receives your Garmin device data. Sync your device to Garmin Connect first. ChatGPT. freddy is an official app in ChatGPT, so you do not need a paid ChatGPT plan to use it with freddy. On a work or school ChatGPT workspace, an admin may need to allow the freddy app first. Set up Garmin in ChatGPT Part 1: Get Garmin into freddy Sync your Garmin device to Garmin Connect. freddy can only read what has reached Garmin Connect. In freddy, open Sources, find Garmin Connect and select Connect. This works on the web and in the app. Sign in to Garmin with the same account that receives your device data. In the app, a "Switch on Historical Data" sheet appears first. History stays off unless that switch is on, so turn it on if you want history, then select Continue to Garmin. Review the data categories and allow the ones you want to share. Back in freddy, let the first import progress and check the dates that have arrived on the Garmin source card. Want more than the recent window? Connecting pulls about 30 days. For older history, use Import account history or Import older Garmin data with your Garmin data export. See Import Garmin account history and FIT files . Part 2: Add freddy to ChatGPT Open the freddy app page in ChatGPT . Select Connect. If ChatGPT shows an install step first, select that. freddy's sign-in page opens. Sign in with your freddy email code if asked, review the Terms, Privacy and Consumer Health Data consent, then approve ChatGPT's access. Enter any sign-in code on freddy's page, never in the chat. Back in ChatGPT, freddy is ready to use. Part 3: Ask your first question Start a new ChatGPT chat. Type @freddy, or select + and pick freddy from the menu. Ask: "What Garmin data is available in my freddy account, and how far back does each type go?" ChatGPT can then show you the metrics and date ranges freddy holds, so you know what to ask next. It may ask you to approve before it reads your data. Review the request and approve it. 6 questions to ask ChatGPT about your Garmin data Start each one with @freddy. Copy these as they are, or change the dates. Sleep: "@freddy How did I sleep last night on my Garmin? Give me duration and stages, and tell me which source you used." Stress: "@freddy Pull my Garmin stress for the last 7 days: daily average, max and qualifier, and time in each stress band. Which days look different from the rest, and what came before them?" Body Battery: "@freddy How much did my Garmin Body Battery charge and drain each day this week? Which days drained the most, and what happened before them?" HRV: "@freddy Show my Garmin overnight HRV for the last 7 days: last night average, 5 minute high and status. Which nights stand out? Give the dates and the source." Workouts: "@freddy Show my last five Garmin workouts with duration, average and max heart rate, and distance, pace and elevation where recorded. Which one was hardest?" Putting it together: "@freddy Compare my Garmin workouts with my sleep and Body Battery charge and drain this week. Do hard days show up the next morning?" Tip: ask ChatGPT to name the source and date range it used. It makes the answer easy to check. On Free, ask about the last 7 days. The full imported history unlocks with Pro. If data is missing Start in Garmin Connect. Check that your device has uploaded the record there. freddy can only read what has reached Garmin Connect. Recent history is absent: check the historical data permission you gave when you connected, and the import status on the Garmin source card. Do not delete the source just to retry a running import. Months or years are missing: connecting pulls about 30 days, and syncing again cannot go further. Use Import account history or Import older Garmin data for older history. New records are missing: use Sync on the Garmin source in freddy after confirming they are in Garmin Connect. ChatGPT says it has no data: check that the dates you asked about are inside your plan's history window, and that ChatGPT is connected to the same freddy account. On Free, your AI's queries cover the last 7 days, and the full imported history unlocks with Pro. The page that opens when you connect shows "Signed in as". If it is the wrong email, select Use another account. ChatGPT keeps asking you to sign in: disconnect freddy in ChatGPT, then connect it again from the freddy app page. More in Understand sync and find missing data , Garmin Connect: connect, sync and troubleshoot and Connect freddy from ChatGPT . Your data, your ChatGPT freddy makes your own Garmin data available to your own AI. You own it. You can pause a source, and if you delete your freddy account, your data is deleted. See Delete your account . To stop ChatGPT's access, disconnect freddy in ChatGPT's settings. You can also revoke it in freddy under Connected AI, which stops access at once. Use Paused on the Garmin source to stop future sync and keep your stored history. Your records in Garmin Connect stay where they are. FAQ Can ChatGPT read my Garmin data? Yes, through freddy. You connect your Garmin account to freddy, and ChatGPT reads that data from freddy once you add the freddy app. ChatGPT does not connect to Garmin directly. Do I need to add a custom connector or paste a URL? No. freddy is an official app in ChatGPT. Open the freddy app page, select Connect and approve. There is no URL to paste and no key to copy. Custom MCP servers are a different route, and you do not need one. Which Garmin account do I use? The one that receives your device data. You sign in to Garmin when you connect and choose which data categories to share. What Garmin data can I ask about? Daily activity, heart rate, sleep, stress and Body Battery where they are recorded and shared. HRV arrives as overnight RMSSD (last night average, 5 minute high and status). Stress comes as a daily average, max, qualifier and time in each stress band. Body Battery comes as daily charged and drained values. Workouts arrive with duration, average and max heart rate, plus distance, calories, pace and elevation where your device recorded them. freddy's sources page also lists steps. What you see depends on what your Garmin device records. How far back does Garmin data go, and what does it cost? Connecting Garmin pulls about 30 days, and Import account history or Import older Garmin data fills in older history from your Garmin export. Free is $0: 1 source, and your AI's queries cover the last 7 days, no card. Pro is $49 a year, and Believer is $249 once for life. Paid plans have no source limit and unlock your full imported history. You do not need a paid ChatGPT plan for freddy. See pricing . Related Garmin Connect: connect, sync and troubleshoot Import Garmin account history and FIT files Connect freddy from ChatGPT Choose and connect a data source Understand sync and find missing data HRV across wearables: how to actually compare what your devices report Recipe: Why am I tired today? Sources freddy connects Pricing Delete your account Using Claude instead? Connect Garmin to Claude freddy is an independent service and is not affiliated with, endorsed by, or sponsored by Garmin or OpenAI.

## Connect Oura to Claude
URL: https://freddy.coach/help/oura-claude
+ You can use your Oura data in Claude by connecting your Oura account to freddy, then adding freddy to Claude from Claude's connector directory. Then ask Claude about your own sleep, readiness, HRV, temperature trends and workouts, and it answers from your real records. The Oura app holds what your ring records. freddy makes that data available to your own Claude, and it stays yours. TL;DR In freddy (web, iPhone or Android app), open Sources, find Oura and choose Connect. Sign in to Oura, review the permissions, and authorize the categories you want to share. Back in freddy, let the first import run and check which dates have arrived. Open freddy in Claude's connector directory and select Connect to Claude. Sign in to freddy, approve, and ask Claude about your Oura data. Free works for this: 1 source, and your assistant can read the last 7 days of your data per question. No card. What you need A freddy account. Sign up free at freddy.coach . An Oura account that holds your ring data. Open the Oura app and let your ring sync first. If your ring is Gen3 or newer, an active Oura Membership. Oura says Gen3 and newer ring owners need one to access their data through other apps ( Oura: The Oura API ). Claude on the web (claude.ai), in the desktop app, or on your phone. You do not need a paid Claude plan to add freddy. On Claude Team and Enterprise, an Owner adds freddy for your organization first. Set up Oura in Claude Part 1: Get Oura into freddy Open the Oura app and let your ring sync. freddy can only read what Oura has processed. In freddy (web, iPhone or Android app), open Sources, find Oura and choose Connect. Sign in to the Oura account that holds your records. You sign in on Oura's own screen. Review the permissions and authorize the categories you want to share. Back in freddy, check the import status and the date coverage on the Oura source card. A connected account does not mean the first history import is complete. New Oura categories may need renewed authorization later. If a category goes missing, check what you granted. Part 2: Add freddy to Claude Open freddy in Claude's connector directory . You can also go to Customize, then Connectors, and search for freddy. Select Connect to Claude. Claude opens a freddy sign-in page. Sign in with the email you use for freddy, then approve. Back in Claude, freddy is ready to use. Prefer your phone? Open the directory link in your mobile browser and connect there. Once connected, freddy also works in the Claude mobile app. Part 3: Ask your first question Start a new Claude conversation. If Claude does not use freddy, select + in the message box, then Connectors, and turn freddy on. Ask: "What Oura data can you see for me this week, and which types are there?" Claude can then show you the metrics and dates it can read, so you know what to ask next. 6 questions to ask Claude about your Oura data Copy these as they are. They fit the 7 days Free can read. On a paid plan, widen the dates. Sleep: "How did I sleep last night on my Oura? Give me duration and stages, and name the source." HRV: "Show my Oura nightly HRV for the last 7 days and tell me which nights stand out. Name the source and dates." Readiness: "How has my Oura readiness looked this week? Which mornings were low, and what came before them?" Temperature: "Show how my Oura temperature deviation moved over the last 7 days. Did anything move away from my usual baseline, and what did my sleep look like then?" Workouts: "List my Oura workouts from the last 7 days with activity, calories, distance and intensity. Which one looks hardest?" Putting it together: "Line up my Oura sleep and readiness with my workouts this week. Do hard days show up the next morning?" Tip: ask Claude to name the source and date range it used. It makes the answer easy to check. A few Oura basics These come from Oura's own support pages, so the numbers mean what Oura says they mean. Readiness is a 0 to 100 score. Oura rates 85 to 100 as Optimal, 70 to 84 Good, 60 to 69 Fair and 0 to 59 Pay Attention. Oura's balance measures compare your last 14 days with your last two months, so a new ring needs up to about two weeks before it knows your averages. ( Oura: Readiness Score , Readiness Contributors ) HRV is measured while you sleep, from five-minute samples through the night. Oura reports a nightly average, and says to compare it with your own trend, not other people's. freddy shows that nightly average in ms (RMSSD). ( Oura: Heart Rate Variability ) Temperature is skin temperature measured at night and shown as a change from your own baseline, in degrees celsius. freddy also shows a trend change. It is not core body temperature. ( Oura: Body Temperature ) Sleep score is a 0 to 100 score built from total sleep, efficiency, restfulness, REM, deep sleep, latency and timing. ( Oura: Sleep Score , Sleep Contributors ) If data is missing Start in the Oura app. Open it, let the ring sync, and check that the sleep session has finished processing. A night or daily score can show up after Oura finishes processing it. Only some categories appear: check the read permissions you granted to freddy. New Oura categories may need renewed authorization. Older data is missing: check the date coverage on the Oura source card before using Re-import all history. What your AI can query is separate from what freddy has stored. Gen3 or newer ring and nothing arrives: check that your Oura Membership is active. New records are missing: use Sync on the Oura source in freddy after confirming they are in the Oura app. Claude says it has no data: check that the dates you asked about are inside your plan's history window, and that Claude is connected to the same freddy account. Free can read the last 7 days of data per question. More in Understand sync and find missing data and Oura: connect, sync and troubleshoot . Your data, your Claude freddy makes your own Oura data available to your own AI. You own it. You can pause a source, and if you delete your freddy account, your data is deleted. See Delete your account . Use Paused on the Oura source to stop future sync and keep your stored history. Delete source erases the stored Oura data from freddy, so read the confirmation first. Your records in Oura stay where they are. You can also revoke freddy's access in Oura's connected-app settings. FAQ Can Claude read my Oura data? Yes, through freddy. You connect your Oura account to freddy, and Claude reads that data from freddy once you add freddy from Claude's connector directory. Claude does not connect to Oura directly. Do I have to give freddy my Oura password? No. You sign in on Oura's own screen when you connect, and then choose which categories to share. What Oura data can I ask about? Daily readiness and sleep scores, sleep sessions and stages, your nightly average HRV (RMSSD, in ms), temperature as a change in celsius from your own baseline plus a trend change, workouts (activity, calories, distance, intensity, start and end time) and heart rate readings. Resting heart rate does not arrive as a bpm number. The closest are your lowest sleeping heart rate and the resting heart rate contributor in your readiness. Stress, resilience and heart-health records come through where your ring, account and permissions support them. What you see depends on your ring and what you authorize. How far back does Oura data go? When you connect, freddy imports your past Oura data, then keeps adding new records as Oura processes them. What your assistant can read per question depends on your plan: Free covers the last 7 days of data, and paid plans cover your full history. What does it cost? Free is $0: 1 source, and your assistant can read the last 7 days of data per question, no card. Pro is $49 a year, and Believer is $249 once for life. Paid plans have no source limit and show your full history. You do not need a paid Claude plan. See pricing . Who owns the data, and can I delete it? You do. freddy makes your own data available to your own AI. If you delete your freddy account, your data is deleted. Steps are in Delete your account . Related Choose and connect a data source Connect freddy to Claude Understand sync and find missing data Oura: connect, sync and troubleshoot HRV across wearables: how to actually compare what your devices report Recipe: Why am I tired today? Sources freddy connects Pricing Delete your account Using ChatGPT instead? Connect Oura to ChatGPT freddy is an independent service and is not affiliated with, endorsed by, or sponsored by Oura or Claude.

## Connect Oura to ChatGPT
URL: https://freddy.coach/help/oura-chatgpt
+ You can use your Oura data in ChatGPT by connecting your Oura account to freddy, then adding the freddy app in ChatGPT. Then ask ChatGPT about your own sleep, readiness, HRV, temperature trends and workouts, and it answers from your real records. The Oura app holds what your ring records. freddy makes that data available to your own ChatGPT, and it stays yours. There is no custom connector to build and no URL to paste. TL;DR In freddy (web, iPhone or Android app), open Sources, find Oura and choose Connect. Sign in to Oura, review the permissions, and authorize the categories you want to share. Back in freddy, let the first import run and check which dates have arrived. Open the freddy app in ChatGPT and select Connect. Sign in to freddy, approve, then type @freddy in any chat and ask about your Oura data. Free works for this: 1 source, and your assistant can read the last 7 days of your data per question. No card. What you need A freddy account. Sign up free at freddy.coach . An Oura account that holds your ring data. Open the Oura app and let your ring sync first. If your ring is Gen3 or newer, an active Oura Membership. Oura says Gen3 and newer ring owners need one to access their data through other apps ( Oura: The Oura API ). ChatGPT. freddy is an official app in ChatGPT, so you do not need a paid ChatGPT plan to use it with freddy. On a work or school ChatGPT workspace, an admin may need to allow the freddy app first. Set up Oura in ChatGPT Part 1: Get Oura into freddy Open the Oura app and let your ring sync. freddy can only read what Oura has processed. In freddy (web, iPhone or Android app), open Sources, find Oura and choose Connect. Sign in to the Oura account that holds your records. You sign in on Oura's own screen. Review the permissions and authorize the categories you want to share. Back in freddy, check the import status and the date coverage on the Oura source card. A connected account does not mean the first history import is complete. New Oura categories may need renewed authorization later. If a category goes missing, check what you granted. Part 2: Add freddy to ChatGPT Open the freddy app page in ChatGPT . Select Connect. If ChatGPT shows an install step first, select that. freddy's sign-in page opens. Sign in with your freddy email code if asked, review the Terms, Privacy and Consumer Health Data consent, then approve ChatGPT's access. Enter any sign-in code on freddy's page, never in the chat. Back in ChatGPT, freddy is ready to use. Part 3: Ask your first question Start a new ChatGPT chat. Type @freddy, or select + and pick freddy from the menu. Ask: "What Oura data can you see for me this week, and which types are there?" ChatGPT can then show you the metrics and dates it can read, so you know what to ask next. It may ask you to approve before it reads your data. Review the request and approve it. 6 questions to ask ChatGPT about your Oura data Start each one with @freddy. Copy these as they are. They fit the 7 days Free can read. On a paid plan, widen the dates. Sleep: "@freddy How did I sleep last night on my Oura? Give me duration and stages, and name the source." HRV: "@freddy Show my Oura nightly HRV for the last 7 days and tell me which nights stand out. Name the source and dates." Readiness: "@freddy How has my Oura readiness looked this week? Which mornings were low, and what came before them?" Temperature: "@freddy Show how my Oura temperature deviation moved over the last 7 days. Did anything move away from my usual baseline, and what did my sleep look like then?" Workouts: "@freddy List my Oura workouts from the last 7 days with activity, calories, distance and intensity. Which one looks hardest?" Putting it together: "@freddy Line up my Oura sleep and readiness with my workouts this week. Do hard days show up the next morning?" Tip: ask ChatGPT to name the source and date range it used. It makes the answer easy to check. A few Oura basics These come from Oura's own support pages, so the numbers mean what Oura says they mean. Readiness is a 0 to 100 score. Oura rates 85 to 100 as Optimal, 70 to 84 Good, 60 to 69 Fair and 0 to 59 Pay Attention. Oura's balance measures compare your last 14 days with your last two months, so a new ring needs up to about two weeks before it knows your averages. ( Oura: Readiness Score , Readiness Contributors ) HRV is measured while you sleep, from five-minute samples through the night. Oura reports a nightly average, and says to compare it with your own trend, not other people's. freddy shows that nightly average in ms (RMSSD). ( Oura: Heart Rate Variability ) Temperature is skin temperature measured at night and shown as a change from your own baseline, in degrees celsius. freddy also shows a trend change. It is not core body temperature. ( Oura: Body Temperature ) Sleep score is a 0 to 100 score built from total sleep, efficiency, restfulness, REM, deep sleep, latency and timing. ( Oura: Sleep Score , Sleep Contributors ) If data is missing Start in the Oura app. Open it, let the ring sync, and check that the sleep session has finished processing. A night or daily score can show up after Oura finishes processing it. Only some categories appear: check the read permissions you granted to freddy. New Oura categories may need renewed authorization. Older data is missing: check the date coverage on the Oura source card before using Re-import all history. What your AI can query is separate from what freddy has stored. Gen3 or newer ring and nothing arrives: check that your Oura Membership is active. New records are missing: use Sync on the Oura source in freddy after confirming they are in the Oura app. ChatGPT says it has no data: check that the dates you asked about are inside your plan's history window, and that ChatGPT is connected to the same freddy account. Free can read the last 7 days of data per question. The page that opens when you connect shows "Signed in as". If it is the wrong email, select Use another account. ChatGPT keeps asking you to sign in: disconnect freddy in ChatGPT, then connect it again from the freddy app page. More in Understand sync and find missing data and Oura: connect, sync and troubleshoot . Your data, your ChatGPT freddy makes your own Oura data available to your own AI. You own it. You can pause a source, and if you delete your freddy account, your data is deleted. See Delete your account . To stop ChatGPT's access, disconnect freddy in ChatGPT's settings. You can also revoke it in freddy under Connected AI, which stops access at once. Use Paused on the Oura source to stop future sync and keep your stored history. Delete source erases the stored Oura data from freddy, so read the confirmation first. Your records in Oura stay where they are. You can also revoke freddy's access in Oura's connected-app settings. FAQ Can ChatGPT read my Oura data? Yes, through freddy. You connect your Oura account to freddy, and ChatGPT reads that data from freddy once you add the freddy app. ChatGPT does not connect to Oura directly. Do I need to add a custom connector or paste a URL? No. freddy is an official app in ChatGPT. Open the freddy app page, select Connect and approve. There is no URL to paste and no key to copy. Custom MCP servers are a different route, and you do not need one. Do I have to give freddy my Oura password? No. You sign in on Oura's own screen when you connect, and then choose which categories to share. What Oura data can I ask about? Daily readiness and sleep scores, sleep sessions and stages, your nightly average HRV (RMSSD, in ms), temperature as a change in celsius from your own baseline plus a trend change, workouts (activity, calories, distance, intensity, start and end time) and heart rate readings. Resting heart rate does not arrive as a bpm number. The closest are your lowest sleeping heart rate and the resting heart rate contributor in your readiness. Stress, resilience and heart-health records come through where your ring, account and permissions support them. What you see depends on your ring and what you authorize. How far back does Oura data go? When you connect, freddy imports your past Oura data, then keeps adding new records as Oura processes them. What your assistant can read per question depends on your plan: Free covers the last 7 days of data, and paid plans cover your full history. What does it cost? Free is $0: 1 source, and your assistant can read the last 7 days of data per question, no card. Pro is $49 a year, and Believer is $249 once for life. Paid plans have no source limit and show your full history. You do not need a paid ChatGPT plan for freddy. See pricing . Who owns the data, and can I delete it? You do. freddy makes your own data available to your own AI. If you delete your freddy account, your data is deleted. Steps are in Delete your account . Related Choose and connect a data source Connect freddy from ChatGPT Understand sync and find missing data Oura: connect, sync and troubleshoot HRV across wearables: how to actually compare what your devices report Recipe: Why am I tired today? Sources freddy connects Pricing Delete your account Using Claude instead? Connect Oura to Claude freddy is an independent service and is not affiliated with, endorsed by, or sponsored by Oura or OpenAI.

## Connect Polar to Claude
URL: https://freddy.coach/help/polar-claude
+ You can use your Polar data in Claude by connecting your Polar account to freddy, then adding freddy to Claude from Claude's connector directory. Then ask Claude about your own sleep, Nightly Recharge, cardio load, heart rate and runs, and it answers from your real records. Polar Flow holds what your Polar watch or tracker records. freddy makes that data available to your own Claude, and it stays yours. TL;DR In freddy, open Sources and set up Polar. Select Connect live sync, sign in to Polar and approve the read permissions you want to share. Select Connect full history with the same Polar account. Sync your Polar device with Polar Flow, then check each connection's progress in freddy. Open freddy in Claude's connector directory and select Connect to Claude. Sign in to freddy, approve, and ask Claude about your Polar data. Free works for this: both Polar connections count as 1 source, with 7 days of history shown. No card. What you need A freddy account. Sign up free at freddy.coach . A Polar Flow account that receives your Polar device data. Claude on the web (claude.ai), in the desktop app, or on your phone. You do not need a paid Claude plan to add freddy. On Claude Team and Enterprise, an Owner adds freddy for your organization first. Set up Polar in Claude Part 1: Get Polar into freddy freddy sets Polar up as a 2-step connection, shown as Polar live sync and Polar history sync in your Sources. They use the same Polar account, and together they count as one source on Free. In freddy, open Sources and choose Set up Polar. This works on the web and in the app. Select Connect live sync. Sign in to Polar and approve the read permissions you want to share. Select Connect full history and use the same Polar account. Sync your Polar device with Polar Flow, then come back to freddy and check each connection's progress. Why two? Live sync adds SleepWise and cardio load, and reaches back roughly 30 days. History sync brings in the older records Polar makes available. Connecting both gives you the broadest coverage, and one does not grant the other. Part 2: Add freddy to Claude Open freddy in Claude's connector directory . You can also go to Customize, then Connectors, and search for freddy. Select Connect to Claude. Claude opens a freddy sign-in page. Sign in with the email you use for freddy, then approve. Back in Claude, start a new conversation (Part 3). Prefer your phone? Open the directory link in your mobile browser and connect there. Once connected, freddy also works in the Claude mobile app. Part 3: Ask your first question Start a new Claude conversation. If Claude does not use freddy, select + in the message box, then Connectors, and turn freddy on. Ask: "What Polar data do you have for me, and how far back does each type go?" Claude can then show you the metrics and date ranges freddy holds, so you know what to ask next. 6 questions to ask Claude about your Polar data Copy these as they are, or change the dates. Sleep: "How did I sleep last night according to Polar? Give me duration and stages, and name the source." Nightly Recharge: "Show my Polar Nightly Recharge for this week. Which nights stand out, and what did my sleep and training look like around them?" Cardio load: "How has my Polar cardio load changed this week? Name the source and the dates." Runs: "List my last five runs with duration, average heart rate and max heart rate. Which one was the hardest?" SleepWise: "What does my Polar SleepWise data say about my alertness this week?" Putting it together: "Line up my Polar training with my sleep, HRV and Nightly Recharge this week. Do hard days show up the next morning?" Tip: ask Claude to name the source and date range it used. It makes the answer easy to check. On Free, ask about the last 7 days. Paid plans show your full history. If data is missing Start at Polar Flow. Make sure your device has synced there. Polar says to sync your watch after you wake up to see last night's sleep, and freddy can only read what has reached Polar Flow. Only some data arrived: open the Polar source in freddy and check both connections. SleepWise and cardio load come through live sync. Older training comes through history sync. No Nightly Recharge: Polar needs continuous heart rate tracking turned on, three nights of wear with three successful sleep measurements, and more than 4 hours of sleep before it shows a Nightly Recharge. Polar's own Nightly Recharge page lists the details. Recent data is missing from both: use Sync in freddy after confirming it is in Polar Flow. Claude says it has no data: check that the dates you asked about are inside your plan's history window, and that Claude is connected to the same freddy account. More in Understand sync and find missing data and Polar: connect, sync and troubleshoot . Your data, your Claude freddy makes your own Polar data available to your own AI. You own it. You can pause a source, and if you delete your freddy account, your data is deleted. See Delete your account . Use Paused on the Polar source to stop future sync and keep your stored history. Delete source erases the stored Polar data from freddy, so read the confirmation first. Your Polar Flow records stay with Polar. FAQ Can Claude read my Polar data? Yes, through freddy. You connect your Polar account to freddy, and Claude reads that data from freddy once you add freddy from Claude's connector directory. Claude does not connect to Polar directly. Why does freddy show two Polar connections? They cover different data. Polar live sync adds SleepWise and cardio load, and its history window is roughly 30 days. Polar history sync brings in older records. Nightly Recharge comes through both, under different field names. Connect both for the fullest picture, using the same Polar account. What Polar data can I ask about? Training sessions, daily activity, sleep, Nightly Recharge and HRV. Runs arrive with duration, average heart rate and max heart rate, and distance comes through live sync. Resting heart rate and cardio load come through live sync only, for the days Polar returns them. What you see depends on what your Polar device records and the permissions you approved. What does it cost? Free is $0: both Polar connections count as 1 source, with 7 days of history shown, no card. Pro is $49 a year, and Believer is $249 once for life. Paid plans have no source limit and show your full history. You do not need a paid Claude plan. See pricing . Who owns the data, and can I delete it? You do. freddy makes your own data available to your own AI. If you delete your freddy account, your data is deleted. Steps are in Delete your account . Related Polar: connect, sync and troubleshoot Connect freddy to Claude Choose and connect a data source Understand sync and find missing data HRV across wearables: how to actually compare what your devices report Recipe: Why am I tired today? Sources freddy connects Pricing Using ChatGPT instead? Connect Polar to ChatGPT freddy is an independent service and is not affiliated with, endorsed by, or sponsored by Polar or Claude. Polar specifics worth knowing. Polar's AccessLink API is the official path Freddy uses; it returns sleep, training sessions, daily activity, Nightly Recharge, Cardio Load, and SleepWise data depending on which Polar device you have. The Vantage V3 and Pacer Pro expose the full set; entry-level trackers (Ignite, Unite) expose sleep and training but not Nightly Recharge. Freddy reads whatever your account exposes; if a metric is missing in Claude, it is usually because your specific Polar device does not record it. If you wear a Polar Verity Sense or H10 chest strap and use Polar Flow's manual entry, those workouts also show up in the API and Freddy will pull them.

## Delete your account
URL: https://freddy.coach/help/delete-account
How to delete your freddy account from the app, the dashboard or by email.
You can delete your freddy account at any time. Deleting your account signs you out and removes your freddy data. It can't be undone. What happens next When you confirm, you're signed out and we remove your freddy data, including: your account record (your email and profile), every connected data source and all of its stored health and training metrics, your personal MCP URL and its token. If you want a copy of your data, export it before you delete. Open Settings on your dashboard and use the export option there. Delete from the freddy app (iOS and Android) Open the freddy app and sign in. Go to the Settings tab. Scroll to Delete account . Type DELETE to confirm, then tap Delete my account and confirm. You're then signed out of the app. Delete from the dashboard (web) Sign in at freddy.coach . Open Settings . In the Delete account section, type DELETE to confirm, then select Delete account . Delete by email If you cannot sign in, email support@freddy.coach from the address on your account and ask us to delete it. We will delete your account and your freddy data. freddy is operated by reThrive Labs LLC. For privacy questions or data requests, see our Privacy Policy .

## Connect freddy to Claude
URL: https://freddy.coach/help/connect-claude
Connect freddy from Claude's connector directory, on claude.ai or in Claude Desktop.
freddy lets Claude answer questions about your sleep, recovery, workouts, and other health data in any conversation. Once connected, Claude reads your data live from your connected sources. It is a short setup. Before you start A freddy account. Sign up free at freddy.coach . At least one data source connected and synced in your freddy dashboard. Garmin, Polar, Suunto, Oura, Withings, Intervals.icu, Hevy, and more all work. Claude.ai or Claude Desktop. On Claude Team and Enterprise, an Owner adds freddy for your organization first. Step 1: Open freddy in Claude's directory Open freddy's page in Claude's connector directory: Open freddy in Claude . You can also find it in Claude: go to Customize, then Connectors, select Discover and search for freddy. On a phone? Open the link in your mobile browser and connect there. Once it's connected, freddy works in the Claude mobile app too. Step 2: Connect to Claude Select Connect to Claude. Claude opens a freddy page: sign in with the email you use for freddy, then approve. You're back in Claude with freddy connected. On Team or Enterprise, if you see Request instead, an Owner needs to add freddy for your organization first. Step 3: Try it Start a new Claude conversation and try one of these. If Claude doesn't use freddy, select + in the message box, then Connectors, and turn freddy on. 1. "How did I sleep last night?" Claude calls list_metrics to discover what sleep data you have, then query_metrics to fetch sleep duration, deep sleep, REM, and HRV for last night. You get a one-line answer plus context. 2. "What's available in my freddy data?" Claude calls list_metrics , which returns every metric freddy has synced for you, the date range each covers, and the source. Useful right after connecting a new device. 3. "Show me my last five workouts and what zones I trained in." Claude calls query_metrics with include_raw: true to fetch per-workout detail, including zone breakdowns and per-set strength data when applicable. What freddy can answer Sleep : duration, stages, HRV, resting HR, sleep efficiency, hypnograms. Recovery : nightly recharge, autonomic balance, breathing rate, and the recovery values your source reports. Training : workouts, durations, sports, calories, average and max heart rate, zones, training load, cardio strain, per-set strength data. Daily activity : steps, calories, distance. Continuous metrics : glucose (Dexcom), continuous heart rate. Body composition : weight, body fat, body water (Withings). Troubleshooting No data found. Make sure a data source is connected and synced in your freddy dashboard . New connections can take a while to backfill. Authentication required. Claude's session with freddy may have expired. In Claude, go to Customize, then Connectors, select freddy and choose Reconnect. freddy shows up twice. If you added freddy before as a custom connector, it keeps working. If you also connect the directory listing, you'll see two. Keep one: under Customize, then Connectors, open the menu on the one marked Custom and select Remove. Free plan limits. The free plan exposes the last 7 days of data and one data source. Upgrade to Pro for $49 per year to unlock full history with no source limit. Cross-device differences. If you have two devices that both measure the same thing (HRV, for example), the numbers will not match. Devices use different methods (RMSSD vs SDNN), windows, and algorithms. freddy returns the source for every metric so Claude can explain the difference. Your privacy freddy reads your health data with OAuth tokens you grant. We do not share or sell your data. See the privacy policy for the full details. Need help? Email support@freddy.coach , or message us in-app at freddy.coach/app/chat .

## Use the freddy CLI
URL: https://freddy.coach/help/freddy-cli
Install the freddy command-line tool to sign in, connect data sources, and query your health data from a terminal or a headless AI agent.
The freddy CLI is a small command-line tool that signs you in, connects your data sources, and queries your health data without the copy-pasted URLs or the shell recipe the older agent guides describe. It is the easiest way to use freddy from a terminal, and it is built for headless AI agents (Claude Code, Cowork, your own bot) that can shell out to a command. It is free to install and open under the MIT license. The CLI wraps the device-flow sign-in, token storage and refresh, the data source connect handoff, and the MCP calls, so you never touch any of that plumbing yourself. Install You can run it on demand with npx , or install it globally: npx @freddy-coach/cli login # run any command on demand # or npm i -g @freddy-coach/cli # then just: freddy <command> It needs Node.js 20 or newer. The examples below assume the global install ( freddy ... ); if you prefer npx , prefix each command with npx @freddy-coach/cli . Sign in freddy login This prints a sign-in URL. Open it in any browser, approve, and the CLI stores your tokens. By default login requests the mcp account:read scopes, which are enough to read your profile and query your data. To also connect, sync, or disconnect data sources, request the broader scope up front: freddy login --scope "mcp account:read connections:write" You can check your status at any time: freddy status It shows whether you are signed in, which host you are signed in to, your scopes, and whether your access token is still valid. Access tokens refresh automatically on the next command, so you rarely need to think about them. Connect a data source List what you can connect: freddy providers Then connect one. This needs the connections:write scope (see Sign in above): freddy connect oura The CLI prints a URL. Open it in your browser, authorize freddy on the source's site, and the CLI waits until the connection lands, then confirms it. From that point freddy syncs that source daily. A few notes: Replace oura with any connectable provider, for example polar , withings , suunto , or intervals-icu . Run freddy providers to list them all; a few are in limited beta, and your freddy dashboard shows what you can connect today. If you signed in without connections:write , connect will tell you to run freddy login --scope "mcp account:read connections:write" first. Re-running login with the broader scope is safe; it does not lose your existing connections. A few sources connect with an API key rather than a browser flow (Hevy, for example). freddy connect does not handle those; connect them from your freddy dashboard instead. Query your data Once a data source is connected and synced, ask for metrics by name over a recent window: freddy query --metrics hrv,sleep_duration --days 7 --metrics is a comma-separated list. --days is a number from 1 to 365. For longer ranges, use --start and --end with YYYY-MM-DD dates. Add --include-raw to get the full provider payloads alongside the summary. To discover what is available to query: freddy metrics # the metrics you can ask for, with date ranges and units freddy profile # your account, plan, and connected data sources Manage connections freddy sync oura # trigger an incremental sync now freddy disconnect oura # disconnect and remove that source's synced data freddy logout # revoke this device and clear local tokens disconnect is destructive: it removes the data freddy synced for that source. logout revokes the CLI's access on the server and deletes the local tokens, so the next command will ask you to sign in again. Machine-readable output (for agents) Every command accepts --json , which prints a single JSON object to stdout instead of human-readable text. This is what you want when an agent or a script drives the CLI: freddy profile --json freddy query --metrics hrv --days 30 --json freddy status --json If you are building an agent integration, the agent quickstart covers the CLI and the underlying REST and MCP surface. Your permanent MCP URL freddy url prints your permanent MCP URL. Unlike an access token, it takes no Authorization header, so it is the right choice for an MCP client that cannot refresh OAuth tokens (many custom and CLI agents). freddy url Add --json for { "mcpUrl": "..." } . Rotate the URL from your dashboard if it leaks. Configuration Two environment variables change where the CLI talks and where it stores tokens: Variable Default What it does FREDDY_BASE_URL https://freddy.coach The freddy host to use. Set it only if you self-host. FREDDY_CONFIG_DIR ~/.freddy Where the CLI stores your tokens. Tokens live in credentials.json inside the config directory. Keep that file private; anyone who can read it can act as you until you freddy logout or rotate. Scopes login requests scopes; each command needs certain ones: Scope Lets you mcp Query your data ( metrics , query ). account:read Read your profile and connections ( profile , status ). connections:write Connect, sync, and disconnect data sources. The default mcp account:read covers reading and querying. Add connections:write when you want to manage data sources. Troubleshooting "Session expired. Run freddy login again." Your refresh token is no longer valid (it expired or was revoked). Run freddy login to sign in fresh. "Missing connections:write" or a scope error on connect/sync/disconnect. You signed in with read-only scopes. Run freddy login --scope "mcp account:read connections:write" , then retry. connect <provider> says the provider uses an API key. A few providers (Hevy, for example) connect with an API key rather than a browser flow. Connect them from the freddy dashboard . "Timed out waiting for the connection." The browser step has a ten-minute window. If you finished it after that, just run freddy connect <provider> again; it will detect the existing connection. Unknown provider. Run freddy providers to see the exact slugs you can use. For more agent-specific errors (audience mismatch, revoking an agent's access), see Troubleshooting agent sign-in . Need help? Email support@freddy.coach , or message us in-app at freddy.coach/app/chat .

## Connect freddy from any AI agent
URL: https://freddy.coach/help/connect-from-agents
Connect freddy to any AI client: standard OAuth with PKCE for most, device flow for headless agents, and your MCP key as a last resort.
Most AI clients need one thing: the server URL https://freddy.coach/mcp. Paste it in and the client works out the sign-in by itself. The rest of this page is for clients that ask for more. For a specific client, see Connect freddy to Claude , Connect freddy from ChatGPT , or Connect freddy from Claude Code . If something does not work, see agent troubleshooting . 1. Standard OAuth, which most clients use freddy speaks OAuth 2.1 with PKCE and dynamic client registration. Cursor, Muse and most connector platforms handle all of it themselves once they have the server URL, so you sign in with a normal browser approval and nothing to copy. In Claude and ChatGPT, add freddy from their directories instead: see Connect freddy to Claude and Connect freddy from ChatGPT. If a setup screen asks you to fill the details in by hand: Server URL: https://freddy.coach/mcp Authorization URL: https://freddy.coach/oauth/authorize Token URL: https://freddy.coach/oauth/token Registration URL: https://freddy.coach/oauth/register PKCE: required, S256 Scopes: mcp account:read connections:write If the screen asks for a client ID and a client secret, register one yourself. Use the exact redirect URL that the setup screen shows you, in place of the example below. curl -sX POST https://freddy.coach/oauth/register \ -H "Content-Type: application/json" \ -d '{"client_name":"My client","redirect_uris":["https://example.com/callback"]}' The response carries client_id and client_secret. Registration is open, so you do not need anything from us first. 2. Device flow, for agents with no browser If your agent runs in a terminal or on a server and cannot open a redirect, freddy also supports device flow (RFC 8628). The easiest path is the freddy CLI: npx @freddy-coach/cli login walks the whole flow for you, including connecting data sources and refreshing tokens. See Use the freddy CLI . If your agent cannot run the CLI, run this once in a terminal and paste the access token wherever your agent expects an MCP bearer token. The recipe also returns a refresh token for renewal. APP=https://freddy.coach CLIENT=$(curl -sX POST $APP/oauth/register \ -H "Content-Type: application/json" \ -d '{"client_name":"My Agent","redirect_uris":["http://localhost"]}' \ | jq -r .client_id) DA=$(curl -sX POST $APP/oauth/device_authorization \ --data-urlencode "client_id=$CLIENT" \ --data-urlencode "scope=mcp account:read connections:write") DEVICE=$(echo "$DA" | jq -r .device_code) URL=$(echo "$DA" | jq -r .verification_uri_complete) echo "Open: $URL" while :; do R=$(curl -sX POST $APP/oauth/token \ --data-urlencode "grant_type=urn:ietf:params:oauth:grant-type:device_code" \ --data-urlencode "client_id=$CLIENT" \ --data-urlencode "device_code=$DEVICE") ERR=$(echo "$R" | jq -r .error) case "$ERR" in authorization_pending|slow_down) sleep 6 ;; null) echo "$R" | jq -r .access_token; break ;; *) echo "stop: $ERR"; break ;; esac done The recipe prints a sign-in URL. Open it in a browser, approve, and the script prints the access token. 3. Your MCP key, as a last resort Some clients speak MCP but cannot refresh the one-hour access token. The symptom is that everything works for about an hour, then every tool stops responding, often shown as an "unknown tool" error, and the client cannot recover on its own. For those clients, use your direct URL with your MCP key. Your direct URL looks like https://freddy.coach/mcp/<your id> . Send the key with every request as an Authorization: Bearer <key> header. Copy both from the Your direct URL card on the AI apps page. Treat the key like a password. It gives full access to your account and can't be narrowed, so prefer OAuth wherever the client supports it. If it ever leaks, rotate it on the AI apps page and the old key stops working at once. Your direct URL stays the same. If a client can only take a URL and can't send a header, the older https://freddy.coach/mcp/<key> form still works for now. What the client can do once signed in freddy exposes seven tools: get_profile, list_metrics and query_metrics read your data. connect_source, sync_source and disconnect_source manage your connected sources. accept_terms records your agreement when a policy is updated. A signed-in client can also read your audit log and manage your account settings over freddy's REST surface, using the scopes you approved. Privacy and revocation Each sign-in produces a grant tied to your freddy account. You can review and revoke connected clients from your dashboard under Connected AI. Revoking a grant invalidates that client's tokens immediately. Need help? Email support@freddy.coach, or message us in the app.

## Your direct URL and MCP key
URL: https://freddy.coach/help/direct-url
Your direct URL is tied to your freddy account. Add it to an AI app, or pair it with your MCP key for an app that can't sign in. Keep the key private.
Every freddy account has a direct URL and an MCP key. The direct URL is tied to your account. Most of the time you only need it. Your direct URL It looks like https://freddy.coach/mcp/ followed by seven letters and numbers. Add it to an AI app as a remote MCP server, then sign in to freddy when the app asks. To add another freddy account of your own to the same AI app, add that account's direct URL as its own connection. Give each one a name that says which account it is. The link does nothing on its own. It only works after you sign in, or with your key. Your MCP key It's for an app or script that can't sign in. Send it with every request as an Authorization: Bearer <key> header, with your direct URL as the server URL. If an app can only take a URL and can't send a header, the older https://freddy.coach/mcp/<key> form still works for now. Where to find them On the web: the AI apps page, under Your direct URL. In the freddy app: the AI tab. The app asks you to confirm it's you before it shows or copies the key. The key stays hidden until you choose Reveal. Copy key works without revealing it. Keep the key private Anyone who has your MCP key can use your freddy account the way your AI app does. They can: read everything freddy stores for the account, including your email address and plan start a sync, or start connecting a new source (only you, signed in to freddy, can finish that) accept updates to freddy's terms disconnect a source, which also deletes the data freddy stored for it So treat it like a password. Add it to your own AI account only, never to an organization or workspace connector. If it ever ends up somewhere it shouldn't, rotate it. Rotate your key Rotate it from the Your direct URL card, on the web or in the app. The old key stops working at once, including in the older URL form. Give the new key to the apps that still need it. Your direct URL stays the same. Apps you connected with sign-in keep working. Signing out of freddy on a device doesn't stop the key. Only rotating does.

## Connect freddy from Claude Code
URL: https://freddy.coach/help/connect-claude-code
Wire freddy into Claude Code (Anthropic's terminal coding agent) so it can answer questions about your sleep, training, and recovery.
Claude Code is Anthropic's terminal-based coding agent. Once freddy is wired in, you can ask Claude Code things like "summarize last week's sleep and HRV" or "show me the last five workouts" right alongside your code work. This guide gives you two paths: Path A, system prompt (no shell recipe): drop a few lines into Claude Code's instructions and let the agent walk the sign-in itself. Easiest if you live in a project that already uses a CLAUDE.md or system prompt. Path B, paste an access token (one shell recipe, one paste): run a five-line script, get a sign-in token, configure Claude Code's MCP once. Most reliable. Pick whichever fits. Both produce the same result. If you would rather query freddy straight from the terminal (or have Claude Code shell out to a command) instead of wiring an MCP server, the freddy CLI does the sign-in, connect, and query for you with npx @freddy-coach/cli . Before you start A freddy account. Sign up free at freddy.coach . Claude Code installed and logged in. At least one data source connected and synced in your freddy dashboard . Path A: Tell Claude Code about the flow Add the following to your Claude Code system prompt (e.g. in ~/.claude/CLAUDE.md for global use, or in your project's CLAUDE.md ): When the user wants to use freddy and there is no freddy MCP yet: 1. POST to https://freddy.coach/oauth/register with JSON {"client_name":"Claude Code","redirect_uris":["http://localhost"]}. Save the returned client_id. 2. POST to https://freddy.coach/oauth/device_authorization with form data client_id=<saved> scope=mcp account:read connections:write (form-encode the spaces in the scope value as %20 or +). Show the user verification_uri_complete and ask them to open it. 3. Poll POST https://freddy.coach/oauth/token with form data grant_type=urn:ietf:params:oauth:grant-type:device_code client_id=<saved> device_code=<from step 2> every 6 seconds. Save the access_token field from the JSON response when one comes back. If the response has error=authorization_pending keep polling; error=expired_token means restart from step 2. 4. Using the saved access token, run: claude mcp add --transport http freddy https://freddy.coach/mcp \ --header "Authorization: Bearer <access_token>" 5. Tell the user freddy is now configured. Ask them to restart Claude Code so the new MCP loads, then in a new session say something like "what's available in my freddy data?". Then in a Claude Code session, say something like "set up freddy and tell me how I slept last week". Claude Code will ask permission to make those HTTP calls, walk the flow, and open the browser sign-in. After you approve in the browser, the agent finishes the setup and starts answering. Path B: Run the script, paste the token Run this once in a terminal. It prints a sign-in URL, you approve in a browser, and it prints an access token: APP=https://freddy.coach CLIENT=$(curl -sX POST $APP/oauth/register \ -H "Content-Type: application/json" \ -d '{"client_name":"Claude Code","redirect_uris":["http://localhost"]}' \ | jq -r .client_id) DA=$(curl -sX POST $APP/oauth/device_authorization \ --data-urlencode "client_id=$CLIENT" \ --data-urlencode "scope=mcp account:read connections:write") DEVICE=$(echo "$DA" | jq -r .device_code) URL=$(echo "$DA" | jq -r .verification_uri_complete) echo "Open: $URL" while :; do R=$(curl -sX POST $APP/oauth/token \ --data-urlencode "grant_type=urn:ietf:params:oauth:grant-type:device_code" \ --data-urlencode "client_id=$CLIENT" \ --data-urlencode "device_code=$DEVICE") ERR=$(echo "$R" | jq -r .error) case "$ERR" in authorization_pending|slow_down) sleep 6 ;; null) echo "$R" | jq -r .access_token; break ;; *) echo "stop: $ERR"; break ;; esac done Copy the freddy_at_... value the script prints. Then: claude mcp add --transport http freddy https://freddy.coach/mcp \ --header "Authorization: Bearer freddy_at_..." Verify: claude mcp list You should see freddy listed. Restart Claude Code, then try a question like "what's available in my freddy data?". Claude Code will call freddy's tools and answer. Token refresh Access tokens expire after a while. The shell recipe also returns a refresh token ( freddy_rt_... ) you can use to mint new access tokens. If your session has been quiet for a while and you start seeing "Authentication required" errors from the freddy tools, run the recipe again to get a fresh access token, then update the MCP config: claude mcp remove freddy claude mcp add --transport http freddy https://freddy.coach/mcp \ --header "Authorization: Bearer <new freddy_at_...>" For now the manual swap is the workaround. Need help? See agent troubleshooting for common errors. If something is still wrong, email support@freddy.coach or message us at freddy.coach/app/chat .

## Connect freddy to Grok Bot
URL: https://freddy.coach/help/connect-grok-bot
Ask your Grok Bot to add freddy as a custom MCP server, approve access, and ask about your sleep, training and heart rate from your Bot.
Grok Bot is xAI's always-on agent. Each Bot works on its own cloud computer, so it can keep going on a task while you are away. This guide adds freddy to your Bot as a custom MCP server, using a message in chat. It is not for the Grok chat at grok.com (see Connect freddy to Grok ) or for the Grok Build terminal agent. Before you start You need a freddy account, at least one connected source, and Grok Bot. To ask about your data right away, connect a source in freddy first. See Choose and connect a data source . 1. Ask your Bot to add freddy You add freddy by asking your Bot in chat. Open a chat and send: “Add a custom MCP server named freddy at https://freddy.coach/mcp” Say “custom MCP server”, so your Bot adds freddy's address and does not look for a catalog plugin. The URL is the same for every freddy account. It contains no personal token. Use exactly https://freddy.coach/mcp, without adding anything to the path. 2. Confirm when it asks Your Bot repeats the name and the URL and asks you to confirm. Check that the URL reads https://freddy.coach/mcp, then confirm. 3. Sign in to freddy When your Bot shows the authorization step, follow it. freddy's authorization page opens in your browser. Sign in with your freddy email code if needed. Review any required Terms, Privacy and Consumer Health Data consent, then approve your Bot's access. Enter your sign-in code on freddy's page, never in the chat. freddy uses OAuth for this connection. You do not need to give your Bot a freddy API key or access token. Grok Bot keeps the sign-in on its own servers and does not store it on the Bot's computer. 4. Ask about your data Mention freddy in a chat and ask: “What data is available in my freddy account?” Your Bot can then use freddy's tools to find your connected metrics and read them for you. Try “How did I sleep last night?” or “Show me my last five runs.” Manage access You can remove the freddy server from your account's connected servers in Grok Bot. You can also revoke your Bot's access in freddy under Connected AI . Revoking it stops your Bot's access at once. If you use Grok Bot on a team, your admin's connector policy can block custom servers. Ask your admin to allow freddy.coach. If setup does not work Check that your Bot has the exact URL https://freddy.coach/mcp and that it is using OAuth. Your Bot reaches the server over the public internet, so a typo in the URL is the most common cause. Do not paste an API key, MCP key or email code into the chat. If the sign-in still fails, contact freddy support with the error text, without sending a code or token.

## Connect freddy to Grok
URL: https://freddy.coach/help/connect-grok
Add freddy as a custom connector at grok.com so Grok can read your sleep, training and heart rate data in any chat.
Grok is xAI's chat assistant at grok.com. You add freddy once as a custom connector, approve the connection, and Grok can answer questions about your sleep, workouts and heart rate in any chat. This guide is for Grok at grok.com. For Grok's always-on agent, see Connect freddy to Grok Bot . Before you start You need a Grok account and a freddy account. To ask about your data right away, connect a source in freddy first. See Choose and connect a data source . 1. Open connectors Go to grok.com/connectors . Select New Connector, then choose Custom. 2. Enter freddy's address Enter this as the MCP server URL: https://freddy.coach/mcp Use the same URL for every freddy account. It contains no personal token. 3. Approve the connection Complete the authentication Grok asks for. Grok sends you to freddy's authorization page. Sign in with your freddy email code if needed. Review any required Terms, Privacy and Consumer Health Data consent, then approve Grok's access. Enter your sign-in code on freddy's page, not in the Grok chat. freddy uses OAuth for this connection. You do not need to give Grok a freddy API key or access token. 4. Ask about your data Start a chat and try: “What data is available in my freddy account?” Grok finds freddy's tools by itself. After that, ask things like “How did I sleep last night?” or “Show me my last five workouts.” If you are on a Business or Enterprise plan On Grok Business and Enterprise, a team admin provisions connectors in the cloud console before they are available to members. If you do not see freddy, ask your admin. Manage access Remove the connector at grok.com/connectors. You can also revoke Grok's OAuth grant in freddy under Connected AI . Using Grok Build instead Grok Build is xAI's terminal coding agent. Run this once: grok mcp add --transport http freddy https://freddy.coach/mcp The first time Grok Build uses freddy, it opens a browser sign-in. Approve it on freddy's authorization page, the same way as above. If setup does not work Check that Grok has the exact URL https://freddy.coach/mcp and that it is using OAuth. Grok reaches freddy over the public internet. Do not paste an API key, MCP key or email code into the chat. If the sign-in still fails, contact freddy support with the error text, without sending a code or token.

## Connect freddy to OpenAI dots
URL: https://freddy.coach/help/connect-dots
Connect freddy once as a ChatGPT plugin and your OpenAI dot can read your sleep, training and heart rate data too.
A dot is OpenAI's always-on agent. It lives inside ChatGPT, has its own cloud computer, and can also be reached from Slack and Teams. Dots use the same plugins as ChatGPT, so freddy only needs to be connected once. We tested freddy with a dot and it works. Before you start Dots are an OpenAI feature, and availability depends on your OpenAI plan and region. See OpenAI's guide to getting started with your dot for the current details. You also need a freddy account and at least one connected source. See Choose and connect a data source . 1. Add freddy as a ChatGPT plugin freddy is an official plugin in ChatGPT. Open the freddy plugin page and connect it. Follow the steps in Connect freddy from ChatGPT if you have not done this before. Sign in with your freddy email code if needed, review any required Terms, Privacy and Consumer Health Data consent, and approve access. 2. Turn freddy on in your dot On mobile, open your dot's profile, then go to Customize, then Plugins. This screen uses the same settings as ChatGPT. Find freddy and connect it if it is not already connected. 3. Ask about your data Message your dot: “What data is available in my freddy account?” Then try “How did I sleep last night?” or “Show me my last five runs.” For something regular, ask your dot to check your sleep and training every morning and tell you what it finds. Manage access Disconnect freddy under Plugins in ChatGPT or in your dot. You can also revoke OpenAI's OAuth grant in freddy under Connected AI . Revoking it stops access at once, for ChatGPT and for your dot. If setup does not work If you do not see Plugins in your dot, check that dots are available for your plan and region. If freddy is missing from the list, open the freddy plugin page from ChatGPT first and connect it there. Do not paste an API key, MCP key or email code into any chat. If the sign-in still fails, contact freddy support with the error text, without sending a code or token.

## Connect freddy to Muse
URL: https://freddy.coach/help/connect-muse
Ask Meta's personal Muse agent to connect to freddy as a Custom Connector, or add freddy to Muse Code in your terminal, then approve OAuth access to your account.
Muse is Meta's personal AI agent. This guide has two parts. The first uses the Muse web app at muse.ai to create a Custom Connector for freddy and sign in with OAuth. The second adds freddy to Muse Code, Meta's terminal coding agent. It does not cover the separate Meta AI app. Muse on the web Before you start You need a Muse account and a freddy account. To ask about your health data right away, connect a source in freddy first. See Choose and connect a data source . 1. Ask Muse to connect freddy Open the Muse web app . In your first message, ask: “Create a connector for freddy using the remote MCP server at https://freddy.coach/mcp. Use OAuth to connect my freddy account.” The URL is the same for every freddy account. It contains no personal token. Use exactly https://freddy.coach/mcp, without adding anything to the path or pasting your MCP key into the chat. 2. Approve the connection Follow the steps Muse presents. When freddy's authorization page opens, sign in with your freddy email code if needed. Review any required Terms, Privacy and Consumer Health Data consent, then approve Muse's access. Enter your sign-in code on freddy's page, not in the Muse chat. freddy uses OAuth for this connection. You do not need to give Muse a freddy API key or access token. 3. Ask about your data After Muse confirms the connection, try: “What data is available in my freddy account?” Muse can then use freddy's tools to find your connected metrics and answer questions about them. If you have not connected a source, add one in freddy before expecting data to appear. Manage access You can disconnect the Custom Connector in Muse under Settings → Connectors. You can also revoke Muse's OAuth grant in freddy under Connected AI . Meta says information Muse already used may remain in its conversations or memory after a connector is disconnected; review Muse's controls if you want to remove that history. Meta does not review Custom Connectors. Review the access request before you approve it. See Meta's guide to how Muse works with Connectors for the current setup and privacy details. If setup does not work Check that Muse has the exact server URL https://freddy.coach/mcp and that it is using OAuth. Do not paste an API key, MCP key or email code into the chat. If the authorization flow still fails, contact freddy support with the error text, without sending a code or token. Muse Code in your terminal Muse Code is Meta's coding agent for the terminal. It is a separate product from the Muse app, so connecting freddy in one does not connect it in the other. Before you start You need Muse Code installed and signed in, and a freddy account with a connected source. If you have not installed it, Meta's one-line installer for macOS and Linux is curl -fsSL https://dev.meta.ai/install.sh | sh . On Windows, use irm https://dev.meta.ai/install.ps1 | iex in PowerShell. Then run muse and sign in. See Meta's Muse Code guide . 1. Add freddy to your settings Muse Code keeps your settings in ~/.config/muse/settings.json . Open that file and add freddy under mcp_servers : { "schema_version": 1, "mcp_servers": { "freddy": { "transport": "streamable_http", "url": "https://freddy.coach/mcp" } } } If the file does not exist yet, create it with the text above. Muse Code needs "schema_version": 1 in the file. If the file already has an mcp_servers block, add the freddy entry inside it and leave the rest as it is. 2. Sign in Run: muse mcp login freddy Muse Code takes you through freddy's sign-in, in your browser or headless. Sign in with your freddy email code if needed, review any required consent, and approve. Muse Code keeps and refreshes the sign-in for you. If you sign in from another terminal, Muse Code picks it up without a restart. 3. Check and ask Start muse and run /mcp to see freddy and its tools in the live list. Then ask: “What data is available in my freddy account?” Manage access Run muse mcp logout freddy to sign out, and remove the freddy entry from your settings file to stop using it. You can also revoke the grant in freddy under Connected AI . If setup does not work If a tool call says you are not signed in, run muse mcp login freddy again. Check that the URL is exactly https://freddy.coach/mcp. Do not paste an API key or MCP key into your settings file when OAuth works.

## Connect freddy to Cursor
URL: https://freddy.coach/help/connect-cursor
Add freddy to Cursor with one line of config and a browser sign-in, then ask about your sleep and training next to your code.
Cursor is the AI code editor. Once freddy is added, you can ask things like “show me the last five workouts” or “how did I sleep this week” without leaving your editor. Cursor handles the OAuth sign-in itself, so there is nothing to paste. Before you start You need a freddy account, at least one connected source, and Cursor installed. See Choose and connect a data source . 1. Add freddy to your MCP config Open ~/.cursor/mcp.json for all projects, or .cursor/mcp.json inside one project. Add: { "mcpServers": { "freddy": { "url": "https://freddy.coach/mcp" } } } If the file already has an mcpServers block, add the freddy entry inside it. The URL is the same for every freddy account and contains no personal token. 2. Sign in Cursor shows freddy on its Customize page. Open freddy there and sign in when Cursor asks. freddy's authorization page opens in your browser. Sign in with your freddy email code if needed, review any required Terms, Privacy and Consumer Health Data consent, and approve. Cursor registers itself with freddy, so you do not enter a client ID or secret. 3. Ask about your data Open Agent chat and ask: “What data is available in my freddy account?” Then try “How did I sleep last night?” Cursor Cloud Agents Cloud Agents keep their own MCP sign-ins on Cursor's servers, separate from the sign-ins on your machine. Signing out on your machine does not sign out a Cloud Agent. To sign out there, open Customize and log out on freddy's Cloud row. Manage access Remove the freddy entry from your config to disconnect. You can also revoke Cursor's OAuth grant in freddy under Connected AI . If setup does not work Check that the URL is exactly https://freddy.coach/mcp. In Cursor, open Output and choose MCP Logs to see what happened. If freddy shows as signed out, sign in again from the Customize page. Do not paste an API key into the config when OAuth works. If it still fails, see agent troubleshooting or contact freddy support with the error text, without sending a code or token.

## Connect freddy from ChatGPT
URL: https://freddy.coach/help/connect-chatgpt
Add the official freddy plugin in ChatGPT, approve access, and ask about your sleep, training and heart rate in any chat.
freddy is an official plugin in ChatGPT. Connect it once and ChatGPT can answer questions about your sleep, training, heart rate and other health data in any chat. You connect it in the ChatGPT app. Before you start You need a freddy account ( sign up free at freddy.coach ) and at least one connected source. See Choose and connect a data source . 1. Open freddy in ChatGPT Open the freddy plugin page in ChatGPT and select Connect. There is no URL to paste and no key to copy. 2. Sign in to freddy freddy's authorization page opens. Sign in with your freddy email code if needed, review any required Terms, Privacy and Consumer Health Data consent, then approve ChatGPT's access. If asked, enter your sign-in code on freddy's page, never in the chat. 3. Ask about your data In any chat, type @freddy or pick freddy from the + menu. Then ask: “What data is available in my freddy account?” After that, try “How did I sleep last night?” or “Show me my last five workouts.” What ChatGPT can do From a chat you can: Ask about your data across all your connected sources. Sync a source. Connect or disconnect a source. To manage your account and sources, use your freddy dashboard . To connect freddy from another AI, see Connect freddy from any AI agent . OpenAI dots A dot uses the plugins you connect in ChatGPT, so freddy is ready in your dot too. See Connect freddy to OpenAI dots . Manage access Disconnect freddy in ChatGPT. You can also revoke ChatGPT's access in freddy under Connected AI . Revoking it stops access at once, for ChatGPT and for your dot. If setup does not work Connected to the wrong freddy account. Look at “Signed in as” on the freddy page that opens when you connect. If it shows the wrong email, select Use another account, sign in with the right one, and approve. Your other account still lists ChatGPT under Connected AI until you remove it there. ChatGPT keeps asking me to sign in. ChatGPT's sign-in with freddy may have expired. Disconnect freddy in ChatGPT, then connect it again from the freddy plugin page. ChatGPT finds no data. Check that a source is connected and synced in your freddy dashboard . A new connection can take a while to bring in your history, so give it some time and check again. Free plan limits. Free includes 1 source and shows 7 days of history. Pro is $49 a year, with full history and no source limit. Believer is $249 once, for lifetime Pro. Data privacy. freddy reads your health data with the access you approve. We do not share or sell it. See the Privacy Policy for the full details. Never paste a key or sign-in code into a chat. If it still fails, contact freddy support with the error text, without sending a code or token. Need help? Email support@freddy.coach , or message us in-app at freddy.coach/app/chat .

## Connect freddy to Codex
URL: https://freddy.coach/help/connect-codex
Add freddy to OpenAI Codex with a short config block and one login command, so Codex can read your sleep and training data.
Codex is OpenAI's coding agent. It runs as a command-line tool, an IDE extension and in the ChatGPT desktop app, and all three read the same MCP config. Add freddy once and each of them can answer questions about your sleep, workouts and heart rate. Before you start You need a freddy account, at least one connected source, and Codex installed. See Choose and connect a data source . 1. Add freddy to your config Open ~/.codex/config.toml and add: [mcp_servers.freddy] url = "https://freddy.coach/mcp" The URL is the same for every freddy account and contains no personal token. 2. Sign in Run: codex mcp login freddy Codex starts the OAuth sign-in with freddy. Follow it, sign in with your freddy email code if needed, review any required Terms, Privacy and Consumer Health Data consent, and approve. 3. Check and ask Run codex mcp list to see freddy. In the Codex terminal UI, type /mcp to see it live. Then ask: “What data is available in my freddy account?” Manage access Remove the [mcp_servers.freddy] block to disconnect. You can also revoke Codex's OAuth grant in freddy under Connected AI . If setup does not work Check that the URL is exactly https://freddy.coach/mcp. If freddy shows as signed out, run codex mcp login freddy again. Do not paste an API key into the config when OAuth works. If it still fails, see agent troubleshooting or contact freddy support with the error text, without sending a code or token.

## Connect freddy to Perplexity
URL: https://freddy.coach/help/connect-perplexity
Add freddy to Perplexity as a custom remote connector, sign in with OAuth, and ask about your sleep and training in any thread.
Perplexity can read your freddy data through a custom remote connector. You add it once, approve the sign-in, and Perplexity can answer questions about your sleep, workouts and heart rate in your threads. Before you start You need a freddy account, at least one connected source, and a Perplexity plan that allows custom connectors. That is Pro, Max or Enterprise. On Enterprise, an admin can control whether members can add their own connectors. 1. Open Plugins In Perplexity, open Plugins in the sidebar. If you do not see Plugins yet, open Customize, then Connectors. Choose Add custom, then Connector, then Add MCP connector, then pick Remote. 2. Fill in the form Name: freddy MCP server URL: https://freddy.coach/mcp Authentication: OAuth 2.0 Transport: Streamable HTTP The URL is the same for every freddy account and contains no personal token. Tick the box that confirms you understand the risks of custom connectors, then select Add. The connector appears on your Plugins page. 3. Approve the connection Select the freddy card to start the sign-in. freddy's authorization page opens. Sign in with your freddy email code if needed, review any required Terms, Privacy and Consumer Health Data consent, and approve. Enter your sign-in code on freddy's page, not in a Perplexity thread. freddy uses OAuth for this connection. You do not need to give Perplexity a freddy API key or access token. 4. Ask about your data Start a thread and ask: “What data is available in my freddy account?” Then try “How did I sleep last night?” Manage access Remove the connector from its card under Plugins in Perplexity (the ⋮ menu). You can also revoke Perplexity's OAuth grant in freddy under Connected AI . If setup does not work Check that the URL is exactly https://freddy.coach/mcp, the authentication is OAuth 2.0 and the transport is Streamable HTTP. Do not paste an API key, MCP key or email code into a thread. If the sign-in still fails, contact freddy support with the error text, without sending a code or token.

## Connect freddy to OpenClaw
URL: https://freddy.coach/help/connect-openclaw
Add freddy to OpenClaw with one command and a browser sign-in, so your agent can read your sleep and training data on a schedule.
OpenClaw is an open source AI agent you run yourself. With freddy added, it can read your sleep, workouts and heart rate, and you can ask it to check them on a schedule. Before you start You need a freddy account, at least one connected source, and OpenClaw installed. See Choose and connect a data source . 1. Add freddy Run: openclaw mcp set freddy '{"url":"https://freddy.coach/mcp","transport":"streamable-http","auth":"oauth"}' The URL is the same for every freddy account and contains no personal token. 2. Sign in Run: openclaw mcp login freddy OpenClaw prints freddy's authorization link. Open it in your browser, sign in with your freddy email code if needed, review any required Terms, Privacy and Consumer Health Data consent, and approve. If the browser runs on another machine, or cannot reach the address OpenClaw printed, copy the code from the redirect and run openclaw mcp login freddy --code <code> . 3. Check and ask Run openclaw mcp status --verbose to see what OpenClaw has saved, or openclaw mcp probe freddy to open a live connection and test it. Then ask your agent: “What data is available in my freddy account?” Manage access Remove freddy from your OpenClaw MCP config to disconnect. You can also revoke OpenClaw's OAuth grant in freddy under Connected AI . If setup does not work Run openclaw mcp doctor freddy --probe and check that the URL is exactly https://freddy.coach/mcp and that auth is oauth . If status says authorization is required, run openclaw mcp login freddy again. Do not paste an API key into the config when OAuth works. If it still fails, see agent troubleshooting or contact freddy support with the error text, without sending a code or token.

## Connect freddy from Gemini
URL: https://freddy.coach/help/connect-gemini
Add freddy as a custom app in Gemini so it can read your health data in any chat. Needs Gemini Spark, US only.
Gemini can read your freddy data through a custom app. You paste one URL, approve the connection, and Gemini can answer questions about your sleep, workouts and recovery in any chat. This one has a narrow door. Gemini custom apps are part of Gemini Spark, and Spark needs a paid Google AI Pro or Ultra subscription. They are also available only in the United States, only in English, and only on a personal Google Account. If that is not your setup, Gemini cannot add freddy today, and the freddy CLI or a client like Claude or ChatGPT will get you there instead. Before you start A Google AI Pro or Ultra subscription, which is what gives you access to Gemini Spark. A personal Google Account. Work and school accounts cannot add custom apps. Keep Activity switched on in your Gemini settings. You are 18 or over, in the United States, and using Gemini in English. A computer. Custom apps cannot be added from the Gemini mobile app, though freddy works there once it is connected. Your freddy MCP URL, which is on your dashboard under Connect freddy to your AI. Add freddy to Gemini On your computer, open gemini.google.com. At the bottom, click Settings & help, then Connected Apps. Under Custom apps for Spark, click Add a custom app. Enter your freddy MCP URL and click Next. freddy registers itself, so you should not need to open Advanced features or enter any credentials. Follow the on-screen instructions. Gemini sends you to freddy to approve the connection. Ask Gemini something like How did I sleep last night? to check it worked. What to expect You only add freddy once, on a computer. After that it works in the Gemini app on your phone as well as on the web. Gemini asks you to confirm before anything that writes, such as starting a sync. Reading your data does not prompt. Google notes that it does not control or secure third-party MCP servers, and that data you send to one is handled under that app's privacy practices. What freddy does with your data is set out in the freddy privacy policy. If something goes wrong If the connection stalls after you approve it, remove the custom app under Connected Apps in Gemini and add it again. If Gemini cannot reach freddy, check that you copied the whole URL, with no trailing space. You can revoke access at any time, either from the Connected apps card on your freddy dashboard or from Connected Apps in Gemini. Using the Gemini CLI instead Gemini CLI is a different product: a developer tool that runs in your terminal, with none of the Spark limits. Run gemini mcp add --transport http freddy with your freddy MCP URL, then restart the CLI.

## Connect freddy to Hermes Agent
URL: https://freddy.coach/help/connect-hermes
Add freddy to Hermes Agent with a short config block and one login command, including a device code option for servers with no browser.
Hermes Agent is an open source agent from Nous Research that you run yourself. With freddy added, it can read your sleep, workouts and heart rate, and you can ask it to check them on a schedule. Before you start You need a freddy account, at least one connected source, and Hermes installed. See Choose and connect a data source . 1. Add freddy to your config Open ~/.hermes/config.yaml and add: mcp_servers: freddy: url: "https://freddy.coach/mcp" auth: oauth If the file already has an mcp_servers block, add the freddy entry inside it. The URL is the same for every freddy account and contains no personal token. 2. Sign in Open a fresh terminal and run: hermes mcp login freddy Hermes prints freddy's authorization link and opens your browser when it can. Sign in with your freddy email code if needed, review any required Terms, Privacy and Consumer Health Data consent, and approve. Run the login from a fresh terminal, not from inside a running Hermes session. When Hermes reloads its config on its own, the sign-in window is short and can time out. On a server with no browser, run hermes mcp login freddy --flow device . Hermes prints a link and a short code. Open the link on any device, enter the code, and approve. 3. Reload and ask In a running Hermes session, type /reload-mcp . Then ask: “What data is available in my freddy account?” Manage access Remove the freddy entry from your config to disconnect. You can also revoke Hermes's OAuth grant in freddy under Connected AI . If setup does not work Check that the URL is exactly https://freddy.coach/mcp and that auth is oauth . If a tool call says you are not signed in, run hermes mcp login freddy again. Do not paste an API key into the config when OAuth works. If it still fails, see agent troubleshooting or contact freddy support with the error text, without sending a code or token.

## Connect freddy to Manus
URL: https://freddy.coach/help/connect-manus
Add freddy to Manus as a custom MCP server so Manus can read your sleep, training and heart rate data.
Manus is a general AI agent. You can add freddy as a custom MCP server, so Manus can answer questions about your sleep, workouts and heart rate. Before you start You need a freddy account, at least one connected source, and a Manus account. See Choose and connect a data source . 1. Open custom MCP servers In Manus, open Connectors, choose Create, then Add MCP by URL. 2. Enter freddy's address Enter these: Name: freddy Server URL: https://freddy.coach/mcp If Manus opens a sign-in, approve on freddy's authorization page after signing in with your email code, and skip the rest of this step. If Manus asks for a credential instead, such as an API key or bearer token, use your direct URL and MCP key instead. Copy both from the Your direct URL card on the AI apps page . Enter the direct URL as the server URL and the key as a bearer token. See Your direct URL and MCP key . Treat the key like a password. It gives full access to your account and cannot be narrowed. 3. Ask about your data Start a task and ask: “What data is available in my freddy account?” Then try “How did I sleep last night?” Manage access Remove the server in Manus to disconnect. If you signed in, revoke the grant in freddy under Connected AI . If you used your MCP key, rotate it on the AI apps page and the old key stops working at once. If setup does not work Check that the server URL is exactly https://freddy.coach/mcp, or your full direct URL if you used the key. Do not paste a key or email code into a Manus chat. If it still fails, contact freddy support with the error text, without sending a code or token.

## Troubleshooting agent sign-in
URL: https://freddy.coach/help/agent-troubleshooting
Common problems and fixes when wiring freddy into Claude Code, Cursor, or other headless AI agents.
If you are setting up freddy with Claude Code, Cursor, or another agent and something is not working, this page covers the common cases. For the basic walkthrough, start with Connect freddy from any AI agent , or skip the manual flow entirely with the freddy CLI . My agent told me to "go to freddy.coach and sign up manually" This is the most common one. Your agent's underlying model does not yet know about freddy's device-flow sign-in, so it falls back to the only thing it knows: pointing you at the sign-up page. You have three options: Use the freddy CLI. If the agent can run a shell command, have it run npx @freddy-coach/cli login and then npx @freddy-coach/cli query --metrics ... --json . The CLI handles sign-in, connecting data sources, and token refresh. See Use the freddy CLI . This is the simplest fix. Tell the agent about the flow. Drop the system-prompt snippet from Connect freddy from Claude Code (Path A) into your agent's instructions or CLAUDE.md . The agent can then walk the flow itself. Run the shell recipe once and paste the access token. This works for any agent that accepts a custom HTTP header for MCP servers. See Path B in the same article. I opened the device URL and it said "expired" The verification URL has a 10-minute window. If you start the flow, then go make coffee, the URL expires before you approve. Just start the flow again. The shell recipe is cheap to re-run; the agent walking it itself can simply call /oauth/device_authorization again. I get "audience_mismatch" This means your token was minted for a different surface than you are calling. Two cases: Token from a different freddy environment. If you have been testing locally against a dev build and copy-pasted a token to production by mistake, you will see this. Use the right environment's token. A ChatGPT-issued token used outside the ChatGPT connector. Tokens issued through the ChatGPT redirect flow are scoped to mcp only and bound to a different audience than tokens from device flow. They cannot be reused against the REST API. Fix: run the device flow yourself and request the scopes you need explicitly, e.g. scope=mcp account:read connections:write . My access token expired Access tokens ( freddy_at_... ) expire after a short time. The agent or shell recipe also returns a refresh token ( freddy_rt_... ) for renewal. To refresh: curl -X POST https://freddy.coach/oauth/token \ --data-urlencode "grant_type=refresh_token" \ --data-urlencode "client_id=<your client_id>" \ --data-urlencode "refresh_token=<freddy_rt_...>" The response carries a fresh access token and a fresh refresh token . Replace your stored refresh token with the new one. If your agent caches the access token in MCP config, you will need to update the Authorization: Bearer ... header with the new value. If your agent has no way to refresh tokens at all, do not fight the expiry. Use your MCP key instead (see the next question). Can I keep using my existing path-token MCP URL? Yes. The URL of the form https://freddy.coach/mcp/<long-token> keeps working for now. The long token in it is your MCP key. It also gets the three new write tools ( connect_source , disconnect_source , sync_source ) automatically. If it leaks, rotate your MCP key on the AI apps page, which stops the old URL at once. You only need the new device flow if you want a token your agent can refresh, or if your agent calls REST endpoints (account management, audit log, billing) in addition to MCP. To keep the key out of the URL, use your direct URL (on the AI apps page) as the MCP server URL and send Authorization: Bearer <key> . My data source connect URL keeps redirecting me to login When an agent connects a data source on your behalf, freddy generates a one-shot /connect/<provider> URL signed for your account. Opening that URL in a browser where you are not signed in will bounce you to login first, then continue the connect flow. If that does not work (you sign in, and the page does not continue to the source's OAuth page), please file a bug. Email the URL, the timestamp, and the agent name to support@freddy.coach , or message us in-app at freddy.coach/app/chat . I get "insufficient_scope" You called an endpoint that needs a scope your token does not have. Common cases: Calling POST /connections/.../sync without connections:write . Reading the audit log without audit:read . Re-run the device flow and pass the scopes you need in the scope parameter, separated by spaces. Available scopes: mcp , account:read , account:write , connections:write , audit:read . I want to revoke an agent's access Visit freddy.coach/app/ai , scroll to "Connected AI", and click revoke on the grant you want to remove. The agent's tokens are immediately invalid. The agent will need to sign in again from scratch the next time you use it. Still stuck? Email support@freddy.coach with the agent name, the exact error, and any logs you have. Or message us in-app at freddy.coach/app/chat . We respond fast.

## Connect freddy to Instinct
URL: https://freddy.coach/help/connect-instinct
How to add freddy to Instinct, the personal agent you text or call, if your account lets you add a custom MCP server.
Instinct is a personal assistant you text or call. Instinct has not published steps for adding your own MCP server. If your account lets you add a custom MCP server, this page shows what to enter. Before you start You need an Instinct account, a freddy account and at least one connected source. See Choose and connect a data source . 1. Find where to add an MCP server Open your Instinct settings and look for a place to add an MCP server or custom connector. If you do not see one, Instinct does not offer it for your account yet. In that case freddy cannot be added to Instinct today. Use another AI client and check back here. 2. Add freddy Add a new server with these details: Name: freddy URL: https://freddy.coach/mcp If Instinct opens freddy's authorization page, sign in with your freddy email code if needed, review any required Terms, Privacy and Consumer Health Data consent, and approve. 3. Ask about your data Text or message Instinct: “What data is available in my freddy account?” Then try “How did I sleep last night?” Manage access Remove the server in Instinct. Then revoke the grant in freddy under Connected AI . If setup does not work Check that the URL is exactly https://freddy.coach/mcp. Do not paste a key or email code into a text message to Instinct. If it still fails, contact freddy support with the error text, without sending a code or token.

## Connect freddy to Hark
URL: https://freddy.coach/help/connect-hark
How to add freddy to Hark, the personal assistant app, if your account lets you add a custom MCP server.
Hark is a personal assistant app for the web, iPhone and Android. Hark has not published steps for adding your own MCP server. If your account lets you add a custom MCP server, this page shows what to enter. Before you start You need a Hark account, a freddy account and at least one connected source. See Choose and connect a data source . 1. Find where to add an MCP server Open Hark's settings and look for a place to add an MCP server or custom connector. If you do not see one, Hark does not offer it for your account yet. In that case freddy cannot be added to Hark today. Use freddy with Claude, ChatGPT, Grok Bot, Muse or another agent in the meantime. We will update this page when Hark publishes steps. 2. Add freddy Add a new server with these details: Name: freddy URL: https://freddy.coach/mcp If Hark opens freddy's authorization page, sign in with your freddy email code if needed, review any required Terms, Privacy and Consumer Health Data consent, and approve. 3. Ask about your data Ask Hark: “What data is available in my freddy account?” Then try “How did I sleep last night?” Manage access Remove the server in Hark. Then revoke the grant in freddy under Connected AI . If setup does not work Check that the URL is exactly https://freddy.coach/mcp. Do not paste a key or email code into a chat. If it still fails, contact freddy support with the error text, without sending a code or token.

## Connect other apps through Apple Health or Health Connect
URL: https://freddy.coach/help/other-bridge-apps
Check an app’s export support, connect the phone bridge and trace missing records from the original app to freddy.
An app can feed freddy without a direct account connection if it writes supported records to Apple Health on iPhone or Health Connect on Android. This guide covers apps beyond the named source catalog. Check whether the route exists Check the app’s own help or settings for Apple Health or Health Connect support on your phone. Google Fit alone is not the same as Health Connect. Check that it can write the data you want. Some integrations only read data into the app. Open the health platform and find an actual record from that app. Check the category, date and source name. A logo, app listing or granted permission does not establish that records have been exported. freddy can only import supported categories that are actually present and readable. Set it up on iPhone In the source app, enable Apple Health sharing and choose the categories it may write. In Apple Health, open the app’s data permissions and verify write access. Inspect an existing exported entry. In the freddy iPhone app, connect Apple Health and allow read access to the matching categories. Keep freddy open during the initial import. Apple Health manual Set it up on Android Enable Health Connect in the source app and grant write access for the chosen categories. In Health Connect, inspect the app permissions and an existing record. Use the same personal phone profile for both apps. In the freddy Android app, connect Health Connect and allow reads for those categories. Review history and background permissions when offered. Health Connect manual Find where data stopped Only in the original app: its export has not reached the phone platform. Check the app’s sharing options, account and sync. In Apple Health or Health Connect, but absent in freddy: check category read permissions, active source status and import progress. In freddy Data, but absent in an AI answer: check the requested dates, source names and your account’s query window. What about old records? The source app decides whether it exports old records. freddy cannot retrieve its private archive through the phone bridge. Health Connect also has a separate historical-read permission. Re-importing does not bypass either limit. What if the app has no dedicated manual or logo? Use the originating app name attached to the platform record to identify it. Recognition and data support are separate: a recognizable name does not mean every metric that app offers can be imported. If a permitted, supported record is present on the phone but missing in freddy, contact support with the app name, category and approximate date. Avoid copies being counted twice A watch workout may pass through several apps before reaching your phone. Compare source, start time and duration before combining totals. Ask your AI client to identify overlapping records and explain which source it uses. Stop sharing Turning off write access stops new exports from the source app. Pausing its source in freddy stops future import there. Clear stored history is separate; read the confirmation for the scope. Revoking a permission does not erase existing copies in every connected app. Apple Health data permissions Health Connect data permissions Related help Choose a source Find missing data and understand sync Contact freddy support

## Import Garmin account history and FIT files
URL: https://freddy.coach/help/garmin-export-import
Add older Garmin records and activity-file detail using the two export import paths in freddy.
The direct Garmin connection can request roughly the last 31 days. Use a Garmin data export for older account history and the separate FIT import for supported activity-file detail. Request your Garmin export Open Garmin Account Data Management and sign in to the account connected to freddy. Choose Export Your Data and request the export. Wait for Garmin’s download email. Download the ZIP and keep the original file. Large archives are easier to handle on a computer. Garmin Account Data Management Import account history Open Garmin Connect in freddy Sources and find the export import options when available for your account. Choose Import account history and select the downloaded ZIP. The browser identifies the supported account-history files. Keep the page open until the upload finishes. After freddy confirms background processing has started, follow the progress on the source card. This route imports supported daily activity, sleep and health history, plus available account-export series such as training and fitness estimates. Some imported series remain distinct from live Garmin series. Completing this step does not mean FIT files have been uploaded. Import FIT activity files Choose Import .FIT activity files from the Garmin export options. Select the archive or files accepted by the import dialog. Review its scan result, selected files and any unreadable-file notice. Start the upload and keep the page open until it finishes. Then check the processing status and the activity dates that have arrived. This separate route adds supported detail from activity recordings. Available fields depend on what the recording device and sensors wrote into each FIT file. A file can contain information that is not yet exposed as a queryable series. Check the result Compare an older date in Data with the Garmin source and import series. Check both account-history and FIT progress if you started both. An upload finishing is not the same as all processing finishing. If new import support is added, the original export can be uploaded again to extract newly supported records. Existing recognized records are handled without requiring you to delete history first. If something goes wrong No import option: check the Garmin source card’s available actions and contact support if the expected feature is missing. Upload interrupted: follow the dialog’s retry or resume action. Keep the page open for the next upload attempt. No supported files found: confirm this is the Garmin export, not a renamed archive from another service. Keep the original structure and file types. Some history is missing: check whether it belongs to account history, FIT detail or a category not included in the export. History and deletion Your AI client’s query window is separate from the history stored in freddy and available in Data or export. Disconnecting Garmin deletes its associated synced and imported data in freddy. Keep your original Garmin ZIP if you may need to restore older history after reconnecting. Garmin connection manual Related help Choose a source Find missing data and understand sync Contact freddy support

## Keep Android health sync running
URL: https://freddy.coach/help/keep-android-sync-running
Check Health Connect permissions, battery settings and foreground import progress when Android updates are delayed.
Health Connect sync runs through the freddy Android app. Android controls background execution, so a connected source does not guarantee an import every few minutes. Check the data path first Open the recording app and let its device sync finish. Open Health Connect and confirm a recent record from that app exists. Open freddy and check whether the record arrives while the app is in the foreground. If the record never reached Health Connect, changing freddy’s battery settings will not fix that first step. Review permissions In Health Connect, allow the recording app to write the chosen category and freddy to read it. Allow background access for freddy when that permission is available and requested. Check the separate history permission if the missing records are old rather than recent. Review Android battery settings Open Android Settings → Apps → freddy and its battery or background-use settings. Allow background activity or choose the unrestricted option if your phone offers one and sync is being delayed. Names differ by manufacturer. On Samsung phones, also check whether freddy has been placed in a sleeping-app list. After force-stopping freddy, open it again before expecting background work. Power-saving modes and manufacturer restrictions can delay tasks even when category permissions are correct. Check the next import Open freddy with an internet connection and let the current import progress. Compare the newest record dates in Health Connect and freddy. Older history can take longer than a small update; keep the app open while you need a foreground import to continue. What these settings do not do Background permission does not make freddy record data itself, force another app to export its archive, or guarantee a fixed sync schedule. It lets freddy use the background opportunities Android provides. Health Connect setup and history permissions Android Health Connect permissions Related help Choose a source Find missing data and understand sync Contact freddy support

## Polar: connect, sync and troubleshoot
URL: https://freddy.coach/help/connect-polar
Bring Polar training, sleep and recovery into the same history as your strength workouts, nutrition and other sources.
Bring Polar training, sleep and recovery into the same history as your strength workouts, nutrition and other sources. Before you connect Use the Polar account that receives your device data in Polar Flow. freddy shows two independent connections: Polar history sync and Polar live sync. Connect Open Sources in freddy Open Polar in Sources and connect Polar history sync. Sign in to Polar and approve the read permissions you want to share. Return to the Polar source and connect Polar live sync with the same Polar account. You can connect the two in either order. Sync your device with Polar Flow, then return to freddy and check each stream’s progress. What reaches freddy Training sessions, daily activity, sleep and Nightly Recharge. Heart rate, heart rate intervals, skin temperature and fitness records where your device and permissions provide them. SleepWise and cardio load through live sync; deeper history through the separate history authorization. History and updates History sync imports the older records Polar makes available and also checks for new data. Live sync adds notifications and data types from Polar’s other service, whose historical window is roughly 30 days. Connecting both gives broader coverage; one authorization does not grant the other. If data is missing One stream is connected: open the Polar source and check both authorizations. Different connected states are possible. SleepWise or cardio load is missing: check live sync. Older training is missing: check history sync and its import progress. Recent data is missing from both: first confirm that it has reached Polar Flow, then use Sync in freddy. Use it with your other sources Line up my Polar sleep and training with my Hevy workouts this week. Show dates and source names, and flag gaps. Pause or disconnect Polar’s two disconnect actions are independent. Disconnecting one removes its authorization and the freddy data attributed to that connection. The other can keep importing. Disconnect both to remove both streams; your Polar Flow records stay with Polar. Provider instructions Polar data access Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Polar data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Oura: connect, sync and troubleshoot
URL: https://freddy.coach/help/oura
Connect Oura sleep, readiness and daily activity so you can compare them with training, food logs and other records in freddy.
Connect Oura sleep, readiness and daily activity so you can compare them with training, food logs and other records in freddy. Before you connect Sync your ring in the Oura app. Oura account or membership restrictions can affect the data its service returns. Connect Open Sources in freddy Open Oura in Sources and choose Connect when available. Sign in to the Oura account containing your records. Review the permissions and authorize the categories you want to share. Return to freddy and inspect the source’s import status and date coverage. What reaches freddy Sleep sessions, stages and daily sleep, readiness and activity summaries. Heart rate, HRV, temperature trends, oxygen saturation and workouts when available. Stress, resilience and heart-health records supported by your ring, account and granted permissions. History and updates freddy imports the history Oura makes available, then receives updates as new records are processed. A night or daily score may appear after Oura finishes processing it. A connected account does not mean the first history import is complete. If data is missing New sleep is missing: open Oura, let the ring sync, and check that the sleep session has finished processing. Only some categories appear: check the read permissions granted to freddy. New Oura categories may need renewed authorization. Older data is missing: inspect the date coverage before using Re-import all history. A query window in your AI client is separate from stored history. Use it with your other sources Compare my Oura sleep with my Garmin training by day this week. Keep missing days visible. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Oura. You can also revoke freddy’s authorization in the source service’s connected-app settings. Provider instructions Oura data and permissions Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Oura data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Withings: connect, sync and troubleshoot
URL: https://freddy.coach/help/withings
Bring your Withings measurements into freddy to read weight, blood pressure, sleep and activity alongside your other sources.
Bring your Withings measurements into freddy to read weight, blood pressure, sleep and activity alongside your other sources. Before you connect Use the Withings account and person profile assigned to your measurements. This matters when several people share a scale or account. Connect Open Sources in freddy Open Withings in Sources and choose Connect when available. Sign in to the Withings account containing your records. Review the permissions and authorize the categories you want to share. Return to freddy and inspect the source’s import status and date coverage. What reaches freddy Weight and body-composition measurements recorded by compatible Withings devices. Blood pressure, pulse, temperature and oxygen measurements where recorded and shared. Daily activity and sleep from devices that support them. History and updates The first import reads available historical measurements and supported activity and sleep records. New records arrive after the device uploads to Withings. A scale measurement assigned to another person will not belong to your connected profile. If data is missing The scale has readings but freddy does not: open Withings and check the person assigned to each measurement. One device is missing: confirm it appears under the connected Withings account and has uploaded recently. Only weight appears: other measures depend on your device, the measurements it recorded and the permissions you granted. Use it with your other sources Show my Withings weight measurements next to my weekly Garmin activity. Include the measurement dates without filling gaps. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Withings. You can also revoke freddy’s authorization in the source service’s connected-app settings. Provider instructions Withings authorization and data permissions Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Withings data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Intervals.icu: connect, sync and troubleshoot
URL: https://freddy.coach/help/intervals-icu
Connect the training and wellness history you keep in Intervals.icu, then compare it with sleep, nutrition and measurements from your other sources.
Connect the training and wellness history you keep in Intervals.icu, then compare it with sleep, nutrition and measurements from your other sources. Before you connect Sign in to your own Intervals.icu athlete account and confirm the activities you want are visible there. freddy uses the account authorization flow; you do not need to paste a personal API key. Connect Open Sources in freddy Open Intervals.icu in Sources and choose Connect when available. Sign in to the Intervals.icu account containing your records. Review the permissions and authorize the categories you want to share. Return to freddy and inspect the source’s import status and date coverage. What reaches freddy Activity summaries and available interval details, including duration, distance, heart rate, power and training load. Wellness records such as sleep, resting heart rate, HRV and weight when present. Source-provided training estimates, kept distinguishable from recorded measurements. History and updates freddy imports available activity and wellness history and keeps the connection updated. Availability can differ for records imported into Intervals.icu from another service. A workout listed in the calendar does not guarantee all its underlying detail is available for export. If data is missing A workout is absent: check its date and origin in Intervals.icu, then use Sync. A detailed chart is missing: the data shared by Intervals.icu can be narrower than the charts shown in its own app. The same ride appears from your device and Intervals.icu: ask your AI client to identify overlapping sessions before calculating totals. Use it with your other sources Compare Intervals.icu training load with my Oura sleep by week. Keep the source’s estimates separate from measurements. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Intervals.icu. You can also revoke freddy’s authorization in the source service’s connected-app settings. Provider instructions Intervals.icu API and account authorization Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Intervals.icu data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Suunto: connect, sync and troubleshoot
URL: https://freddy.coach/help/suunto
Bring Suunto workouts and daily records together with your sleep, strength and nutrition sources in freddy.
Bring Suunto workouts and daily records together with your sleep, strength and nutrition sources in freddy. Before you connect Sync your watch to the Suunto app and use that same Suunto account when connecting freddy. Connect Open Sources in freddy Open Suunto in Sources and choose Connect when available. Sign in to the Suunto account containing your records. Review the permissions and authorize the categories you want to share. Return to freddy and inspect the source’s import status and date coverage. What reaches freddy Recorded workouts, including available distance, duration, heart rate, pace and route information. Daily activity, sleep and recovery records exposed for your device. Additional workout measurements where the Suunto record includes them. History and updates The initial import requests the history Suunto makes available for each data type. Training and daily health records may cover different dates. New records become available after the watch and Suunto app have synced. If data is missing A finished workout is missing: confirm it is saved in the Suunto app and the app has an internet connection. Sleep or recovery is absent: check whether your watch recorded that category and whether it is visible in Suunto. An old workout has fewer fields: freddy can only import the fields present in the shared record, not reconstruct missing sensor data. Use it with your other sources Put my Suunto runs next to my Hevy sessions this week. Show each workout once after checking for overlap. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Suunto. You can also revoke freddy’s authorization in the source service’s connected-app settings. Provider instructions Suunto connected data services Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Suunto data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Concept2: connect, sync and troubleshoot
URL: https://freddy.coach/help/concept2
Connect your Concept2 Online Logbook to bring RowErg, SkiErg and BikeErg sessions alongside the rest of your training.
Connect your Concept2 Online Logbook to bring RowErg, SkiErg and BikeErg sessions alongside the rest of your training. Before you connect Your session must be in the Concept2 Online Logbook. If it is only on the performance monitor, first transfer it using ErgData or your usual Logbook workflow. Connect Open Sources in freddy Open Concept2 in Sources and choose Connect when available. Sign in to the Concept2 account containing your records. Review the permissions and authorize the categories you want to share. Return to freddy and inspect the source’s import status and date coverage. What reaches freddy Online Logbook results with dates, workout type, distance and duration. Pace, stroke rate and heart rate where the saved result includes them. Result summaries rather than a guaranteed stroke-by-stroke recording. History and updates freddy reads the available results in your Online Logbook and imports new saved results. Connecting freddy does not pair it with the monitor or start a Bluetooth recording. If data is missing The monitor shows a workout but freddy does not: check that ErgData has sent it to the Online Logbook. Heart rate is absent: check that heart rate was captured and included in the uploaded result. A result has the wrong date or distance: correct it in the Logbook, then request Sync in freddy. Use it with your other sources Compare my Concept2 rowing time with my Garmin running time this week, using saved workout durations. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Concept2. You can also revoke freddy’s authorization in the source service’s connected-app settings. Provider instructions Concept2 Logbook data and authorization Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Concept2 data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Wahoo: connect, sync and troubleshoot
URL: https://freddy.coach/help/wahoo
Connect Wahoo to bring saved rides and workouts into the same history as your recovery, strength and nutrition data.
Connect Wahoo to bring saved rides and workouts into the same history as your recovery, strength and nutrition data. Before you connect Sync your ELEMNT or other Wahoo recording app with your Wahoo account. Connecting freddy reads saved cloud records; it does not connect directly to a KICKR or heart-rate sensor. Connect Open Sources in freddy Open Wahoo in Sources and choose Connect when available. Sign in to the Wahoo account containing your records. Review the permissions and authorize the categories you want to share. Return to freddy and inspect the source’s import status and date coverage. What reaches freddy Saved workout summaries: time, distance, energy and workout type. Heart rate, power, cadence and related summary metrics when present. Available workout metadata; a summary or FIT reference does not guarantee every per-second field is imported. History and updates freddy imports the workouts Wahoo exposes for your account and picks up new saved records. A ride must finish uploading to Wahoo before it can arrive in freddy. If data is missing A ride is still on your bike computer: finish the Wahoo sync first. Power or heart rate is missing: check the original workout for that sensor’s recording. The same ride also arrives from another training service: compare start time and duration before adding the two sources together. Use it with your other sources Line up my Wahoo rides with my Oura sleep this week. Show ride duration and the preceding night separately. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Wahoo. You can also revoke freddy’s authorization in the source service’s connected-app settings. Provider instructions Wahoo support and workout help Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Wahoo data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Ultrahuman: connect, sync and troubleshoot
URL: https://freddy.coach/help/ultrahuman
Connect Ultrahuman to read your ring’s sleep, movement and recovery records alongside training and other measurements.
Connect Ultrahuman to read your ring’s sleep, movement and recovery records alongside training and other measurements. Before you connect Sync your device in the Ultrahuman app. Ring and glucose features depend on the products and services present in your account. Connect Open Sources in freddy Open Ultrahuman in Sources and choose Connect when available. Sign in to Ultrahuman. If asked for a six-character code, get a fresh one in the Ultrahuman app: Profile → Access Code at the bottom. Review the permissions and authorize the categories you want to share. Return to freddy and inspect the source’s import status and date coverage. What reaches freddy Sleep duration, stages and available sleep summaries. Movement, recovery, resting heart rate, HRV and temperature records where supplied. Glucose records if your Ultrahuman account includes them and the authorized service provides them. History and updates freddy checks Ultrahuman periodically and imports the available daily history. A day’s record may change as sleep and movement processing completes. The first connection and a fully processed day are separate steps. If data is missing Today looks incomplete: check the date and processing state in Ultrahuman, then let the next sync run. Glucose does not appear: a ring connection by itself does not create glucose records. Check the records and permissions in your account. A category stopped updating: compare its latest date in Ultrahuman and freddy before reconnecting. Use it with your other sources Show Ultrahuman sleep duration alongside my Hevy workout dates. Identify missing days instead of estimating them. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Ultrahuman. You can also revoke freddy’s authorization in the source service’s connected-app settings. Provider instructions Ultrahuman support Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Ultrahuman data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Hevy: connect, sync and troubleshoot
URL: https://freddy.coach/help/hevy
Bring your Hevy exercise log into freddy so sets, reps and weights can sit beside endurance sessions and sleep records.
Bring your Hevy exercise log into freddy so sets, reps and weights can sit beside endurance sessions and sleep records. Before you connect Hevy API access requires Hevy Pro. Your freddy account and your Hevy subscription are separate. Connect Open Sources in freddy Open Hevy’s Developer settings and copy your API key. In freddy Sources, choose Hevy and Connect. Paste the key into the API Key field and save. Return to the source and check the import progress, then compare a completed workout with your Hevy log. What reaches freddy Logged workouts, exercise names and detailed sets. Reps, load and available set details from the workout record. Body measurements where available in your Hevy account. History and updates freddy imports your available workout history and checks periodically for new or edited sessions. Save the session in Hevy first. A routine template is not the same as a completed workout. If data is missing The key is rejected: confirm Hevy Pro access and copy the current key without extra spaces. A session is missing: finish and save it in Hevy, then use Sync in freddy. A key stopped working after replacement: update the connection with the new key. Do not send API keys to your AI client or support chat. Use it with your other sources List my Hevy exercises, sets and reps this week beside my Garmin runs. Check for overlapping workout records. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Hevy. You can also revoke the key or API client in the source service’s settings. Provider instructions Find your Hevy API key Hevy API access Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Hevy data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Lyfta: connect, sync and troubleshoot
URL: https://freddy.coach/help/lyfta
Connect Lyfta workout logs so you can compare strength sessions and recorded progress with sleep and other training.
Connect Lyfta workout logs so you can compare strength sessions and recorded progress with sleep and other training. Before you connect Anyone on freddy can connect Lyfta. You’ll need an API key, which you create on Lyfta’s API Access page. Connect Open Sources in freddy Open Lyfta’s API Access page and generate a key. Copy it when it is shown. Open the Lyfta source in freddy, choose Connect, and paste the key into API Key. Save the connection, check the import status, and compare one completed workout with Lyfta. What reaches freddy Saved workouts and the exercises and sets within them. Recorded reps, load and available workout summaries. Available progress records, such as personal records or estimated one-rep max, with their source labels. History and updates The initial import reads available Lyfta history; periodic sync brings in new records. Keep recorded weights and units attached to the source. Estimated one-rep max is a source estimate, not a measured lift. If data is missing A key fails: confirm it has not been revoked and copy it from the correct Lyfta account. A weight looks different: check the original entry’s units and exercise variant before comparing records. Use it with your other sources Compare my Lyfta strength sessions with Oura sleep by date. Keep recorded lifts separate from estimated one-rep max. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Lyfta. You can also revoke the key or API client in the source service’s settings. Provider instructions Lyfta API Access Lyfta API documentation Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Lyfta data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Runalyze: connect, sync and troubleshoot
URL: https://freddy.coach/help/runalyze
Bring Runalyze activities and wellness records into freddy to compare your training analysis with sleep, strength and body measurements.
Bring Runalyze activities and wellness records into freddy to compare your training analysis with sleep, strength and body measurements. Before you connect Anyone on freddy can connect Runalyze. Runalyze’s read API needs a Supporter or Premium account and a personal token with read access. Connect Open Sources in freddy Open Runalyze Settings → Personal API. Enable read access and create a token with an expiry that suits your use. Open the Runalyze source in freddy Sources, choose Connect, and paste the token into API Key. Save, allow the history import to progress, and compare a recent activity with Runalyze. What reaches freddy Activity summaries and available training metrics. Wellness records such as sleep, HRV, weight or blood pressure when recorded. Runalyze estimates and scores where supplied, distinct from the underlying measurements. History and updates freddy imports available activity and wellness history and checks for updates periodically. The data exposed through Runalyze’s API may differ from a full interactive chart in Runalyze. If data is missing Authorization fails: check read access, token expiry and the Runalyze subscription required for API reads. A computed metric differs from another source: keep Runalyze’s calculation identified instead of treating it as the same measurement. Use it with your other sources Compare my Runalyze weekly activity with my Withings measurements. Name each source and retain missing dates. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Runalyze. You can also revoke the key or API client in the source service’s settings. Provider instructions Runalyze Personal API settings Runalyze API requirements Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Runalyze data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Airthings: connect, sync and troubleshoot
URL: https://freddy.coach/help/airthings-setup
Bring readings from your Airthings monitors alongside sleep and other health records, with each room or device kept identifiable.
Bring readings from your Airthings monitors alongside sleep and other health records, with each room or device kept identifiable. Before you connect Use an Airthings consumer account with devices reporting to its cloud. This connection needs a Client ID and Client Secret from a Client Credentials API client. Connect Open Sources in freddy Open the Airthings Consumer API dashboard and its My Applications area. Register an app with the Client Credentials flow. Copy the Client ID and Client Secret. In freddy Sources, choose Airthings and paste them into the matching fields. Save the connection and compare each device’s latest reading with the Airthings app. What reaches freddy Radon, CO₂, VOCs and particulate matter supported by your monitor. Temperature, humidity, air pressure and available device battery information. Timestamped readings from each connected device, plus supported historical CSV records you upload. History and updates The automatic connection reads current device values and builds history forward. For earlier readings, export an IAQ CSV from Airthings, then choose Upload history (.csv) in freddy. Select the correct device, confirm the timezone in which it recorded, and upload the CSV. If data is missing Credentials are rejected: check the flow is Client Credentials and the ID and secret are in the correct fields. A room stopped updating: confirm that monitor is reporting to the Airthings cloud before using Sync. Imported times look shifted: check the timezone selected for the CSV. Do not import another room’s CSV under the wrong device. Use it with your other sources Align my bedroom Airthings CO₂ readings with my Oura sleep times. Show coverage and avoid inferring a cause from the comparison. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Airthings. You can also revoke the key or API client in the source service’s settings. Provider instructions Airthings Client Credentials setup Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Airthings data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Connect WHOOP
URL: https://freddy.coach/help/whoop
Connect WHOOP to freddy and explore sleep, recovery, strain, and workouts alongside your other sources. Setup steps, integration mockups, and troubleshooting.
Bring WHOOP sleep, recovery, HRV, strain, and workouts together with your other health and fitness sources. Compare a night’s sleep with the run you recorded on Garmin or the strength session you logged in Hevy, then explore the pattern through your preferred AI assistant. You connect your own WHOOP account and choose which AI clients can access freddy. WHOOP access is read-only and revocable. Illustrative mockup of the connection flow. Layout may differ between web and mobile. Connect your WHOOP account Start with your WHOOP login and a freddy account. Open Sources in your browser and sign in if prompted. Find WHOOP in the source list and choose Connect. Sign in on WHOOP’s authorization page with the account you use in the WHOOP app. Review the requested access and authorize freddy. Your WHOOP password stays with WHOOP. Return to freddy. Open the WHOOP source card to follow the initial import and check its connection status. You do not need a WHOOP developer account, API key, or manual data export. What freddy reads from WHOOP Available records depend on your WHOOP history, permissions, and whether WHOOP has finished processing them. Sleep: start and end times, sleep stages, time awake, sleep performance, consistency, and respiratory rate. Recovery: recovery score, heart rate variability (HRV), resting heart rate, and blood oxygen or skin temperature when available. Strain: physiological-cycle strain, energy expenditure, and average and maximum heart rate. Workouts: activity type, duration, strain, heart rate, and time in heart rate zones. Profile and body measurements: basic profile details, height, weight, and maximum heart rate when available. These are WHOOP’s records and scores. WHOOP scores stay attributed to WHOOP when you compare them with another source. Put WHOOP alongside your other sources Connect the accounts you want to compare in Sources. For example, use WHOOP for sleep and recovery, Garmin for running or cycling, and Hevy for sets, reps, and strength sessions. Your freddy plan determines how many sources and how much history your AI can query. Then connect freddy to your AI assistant using the AI setup guides and ask a question across the accounts you have linked. Illustrative conversation with synthetic data. The response and layout depend on your AI client, question, and available records. Try asking: Compare my WHOOP sleep and recovery with my Garmin training and Hevy workouts over the last two weeks. Name each source, flag missing days, and check for duplicate workouts before calculating totals. How have my WHOOP HRV and resting heart rate changed over the last month? Compare each with its own recent baseline. If you record one workout in two apps, ask your assistant to identify the overlap rather than add both records. Keep each source’s recovery and HRV measurements separate when comparing them. Sync and history freddy imports the WHOOP history available through the connection, then keeps it updated automatically. You can check progress or request a sync from the WHOOP source card. Larger history imports can take longer than a routine update. New sleep and recovery records may arrive at different times. A record can appear before WHOOP has finished calculating its score. If a value is missing, first check whether it is available in the WHOOP app. Pause or disconnect Pause WHOOP from its source card to stop syncing while keeping the imported data. Resume it when you want updates again. To remove the connection, choose Disconnect in the WHOOP source card and review the confirmation. Disconnecting removes the WHOOP data held in freddy; it does not delete the original records in your WHOOP account. Your AI client has its own data policies. See freddy’s privacy policy for how freddy handles connected data. If something is missing WHOOP asks you to reconnect: use the reconnect action on the source card and authorize the same WHOOP account again. Today’s data is missing: check that your WHOOP app has received and processed it, then run a sync in freddy. Older data is missing: check the import status and your plan’s query window. Re-import history from the source card if the import itself is incomplete. You see a waitlist message: use the waitlist option shown. Still stuck? Message support with the error text and the step where it happened. Do not send your password or access tokens. Reference: WHOOP’s data documentation . More source guides Choose a source Find missing data and understand sync

## Dexcom: connect, sync and troubleshoot
URL: https://freddy.coach/help/dexcom
The Dexcom connection is in alpha in freddy. When access is enabled, it brings historical glucose records into the same timeline as your other sources.
The Dexcom connection is in alpha in freddy. When access is enabled, it brings historical glucose records into the same timeline as your other sources. Before you connect Check the Dexcom source card for access. A listing in the catalog is not confirmation that the alpha is enabled for your account. Use the Dexcom account associated with your recorded data. Connect Open Sources in freddy Open Dexcom in Sources. If access is restricted, use the access or waitlist option shown. When Connect is enabled, sign in through Dexcom and review the requested data permissions. Return to freddy and inspect the dates of the imported records. What reaches freddy Historical glucose readings exposed by the Dexcom data service. Available device, calibration and event records included in the authorized data. The provider’s timestamps and units, so readings can be compared with other dated records. History and updates Dexcom’s shared data can be delayed. This connection is for reviewing historical records, not real-time glucose alerts or treatment decisions. Keep using your Dexcom device and app for their intended monitoring functions. If data is missing Connect is unavailable: alpha access is restricted; reconnecting another source will not enable it. Recent readings are absent: compare timestamps and allow for the Dexcom service’s delay. Glucose units differ across views: retain the units with each value before making comparisons. Use it with your other sources When Dexcom is connected, place historical readings next to my logged meal times. Show units and any gaps in either source. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Dexcom. You can also revoke freddy’s authorization in the source service’s connected-app settings. Provider instructions Dexcom developer data documentation Related help Choose a source Find missing data and understand sync Contact freddy support

## Garmin Connect: connect, sync and troubleshoot
URL: https://freddy.coach/help/garmin-setup
Connect Garmin workouts, sleep and daily records, then add older account history or FIT activity detail from your Garmin export.
Connect Garmin workouts, sleep and daily records, then add older account history or FIT activity detail from your Garmin export. Before you connect Sync your device to Garmin Connect and use the same Garmin account to authorize freddy. Review the historical-data permission as well as the ongoing data categories. Connect Open Sources in freddy Open Garmin Connect in freddy Sources and choose Connect. Sign in to Garmin, review the data categories and approve historical data access if you want the initial history import. Return to freddy. Let the initial requests progress and check the dates that have arrived. For older history or FIT detail, open the Garmin export options on the source card when available. What reaches freddy Daily activity, heart rate, sleep, stress and Body Battery where recorded and shared. Workout summaries and supported activity detail. Available body and health measurements from Garmin, plus supported records from your account export. History and updates The direct Garmin connection can request about a month of history. Older history comes through your Garmin export. Account history and .FIT activity files have separate import options; completing one does not mean the other has finished. If data is missing New records are missing: check that your device uploaded them to Garmin Connect. Recent history is absent: check the historical-data consent and the source’s import status. Avoid disconnecting merely to retry a running import. Months or years are missing: use the archive import guide; repeating the direct sync cannot extend Garmin’s historical window. Use it with your other sources Compare my Garmin runs with my Hevy sessions and WHOOP sleep this week. Identify overlapping workouts before calculating totals. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Garmin Connect. You can also revoke freddy’s authorization in the source service’s connected-app settings. Provider instructions Garmin account data export Import Garmin account history and FIT files Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Garmin data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Apple Health: connect and manage your sources
URL: https://freddy.coach/help/connect-apple-health
Use the freddy iPhone app to bring records from Apple Health into freddy, including Apple Watch data and records written by other apps.
Use the freddy iPhone app to bring records from Apple Health into freddy, including Apple Watch data and records written by other apps. Before you connect Install freddy on the iPhone containing your Health records and sign in to your freddy account. A desktop browser can show uploaded records, but cannot read your iPhone’s Health database. Connect Open freddy Sources after installing the phone app freddy for iPhone freddy for Android Open freddy on iPhone, go to Sources and connect Apple Health. Review the iOS Health permission sheet and allow the categories you want freddy to read. Keep freddy open during the initial import and inspect its progress. For another app’s records, also enable that app’s permission to write those categories to Apple Health. Open the Apple Health source in freddy to review the apps and devices contributing records. What reaches freddy Supported activity, workouts, sleep, heart-rate and related records that you allow freddy to read. Supported body measurements, nutrition and other health categories present in Apple Health. The source app or device attached to imported records, so you can inspect where they came from. History and updates freddy can import supported records already present on the phone and permitted by iOS. Older records that never reached Apple Health cannot be recovered by a freddy re-import. Background updates depend on iOS scheduling; opening freddy lets it resume work in the foreground. If data is missing An app appears in Health but has no records: permission alone is not a data transfer. Open a category in Apple Health and inspect its entries and source. Records are in Apple Health but absent in freddy: check freddy’s read permissions and whether the source is active in freddy. Updates are delayed: open the source app first, then freddy, and check progress. Do not assume a background task has run at a fixed time. Use it with your other sources Compare my Apple Health sleep with my Hevy workout dates. Show the app or device that supplied each sleep record. Pause or disconnect Revoke freddy’s Health read permissions to stop access from the phone. In freddy, pausing a contributing source stops future import; Clear stored history is a separate deletion action. Read its confirmation for the scope. Removing permission does not by itself delete records already stored in freddy or Apple Health. Provider instructions Manage Apple Health data and sources Other apps through Apple Health or Health Connect Related help Choose a source Find missing data and understand sync Contact freddy support

## Connect Apple Health to Claude
URL: https://freddy.coach/help/apple-health-claude
Use your Apple Health data in Claude. Install freddy on your iPhone, add freddy from Claude's connector directory, then ask about your sleep, HRV and workouts.
You can use your Apple Health data in Claude by installing the freddy iPhone app, letting it read Apple Health, and adding freddy to Claude from Claude's connector directory. Then ask Claude about your own sleep, HRV, resting heart rate and workouts, and it answers from your real records. Apple Health holds the records your iPhone, your Apple Watch and many of your other apps write to it. freddy makes that data available to your own Claude, and it stays yours. TL;DR Install freddy on the iPhone that holds your Health records and sign in. In freddy, open Sources, tap Connect on the Apple Health row and allow the categories you want shared. Keep freddy open and your iPhone unlocked until the import finishes. Open freddy in Claude's connector directory and select Connect to Claude. Sign in to freddy, approve, and ask Claude about your data. Free works for this: 1 active source and 7 days of history shown. No card. What you need A freddy account. Sign up free at freddy.coach . An iPhone with your Apple Health records, and the freddy iPhone app . A desktop browser can show records that were already uploaded, but it cannot read the Health database on your iPhone. Claude on the web (claude.ai), in the desktop app, or on your phone. You do not need a paid Claude plan to add freddy. On Claude Team and Enterprise, an Owner adds freddy for your organization first. Set up Apple Health in Claude Part 1: Get Apple Health into freddy Install freddy on your iPhone and sign in with your freddy account. In the app, open the Sources tab, find the Apple Health row and tap Connect. You can also tap + Add source on Home. Your iPhone then shows Apple's own Health permission screen. Allow the categories you want freddy to read, such as sleep, heart rate and workouts. Keep freddy open and your iPhone unlocked during the import. Locking the phone pauses Health access, and your progress is saved. The Apple Health source shows the import while it runs. Open the Apple Health source in freddy to see which apps and devices are contributing records. On Free, freddy shows the last 7 days of history. Paid plans show your full history. Using another app that writes to Apple Health? Turn on that app's permission to write those categories to Apple Health first. Other apps through Apple Health or Health Connect walks through it. Part 2: Add freddy to Claude Open freddy in Claude's connector directory . You can also go to Customize, then Connectors, select Discover and search for freddy. Select Connect to Claude. Claude opens a freddy sign-in page. Sign in with the email you use for freddy, then approve. Back in Claude, start a new conversation (Part 3). Prefer your phone? Open the directory link in your mobile browser and connect there. Once connected, freddy also works in the Claude mobile app. Part 3: Ask your first question Start a new Claude conversation. If Claude does not use freddy, select + in the message box, then Connectors, and turn freddy on. Ask: "What Apple Health data do you have for me, and how far back does it go?" Claude can then show you the metrics and date range freddy holds, so you know what to ask next. 6 questions to ask Claude about your Apple Health data Copy these as they are, or change the dates. Sleep: "How did I sleep last night? Give me duration, stages if you have them, and which app or device recorded it." Sleep trend: "Compare my sleep this week with the last four weeks. What changed?" HRV: "Show my HRV over the last 7 days and tell me which days stand out. Name the source for each reading." Apple Health reports HRV as SDNN, so compare it with your own baseline and not with the RMSSD that other devices report. Resting heart rate: "How has my resting heart rate moved this week? Did it change when my training changed?" Workouts: "List my last five workouts with duration, heart rate and which device logged each one." Putting it together: "Line up my workouts with my sleep and resting heart rate this month. Do hard days show up the next morning?" Tip: ask Claude to name the source and date range it used. It makes the answer easy to check. On Free, ask about the last 7 days. If data is missing It is in Apple Health but not in freddy: check freddy's read permissions in iOS and that the source is active in freddy. The import stopped part way: unlock your iPhone, keep freddy open and tap Resume or Continue if you see it. Progress is saved. Updates look late: open the app that recorded the data first, then open freddy and check its progress. Background updates follow iOS scheduling. Claude says it has no data: check that the dates you asked about are inside your plan's history window, and that Claude is connected to the same freddy account. More in Understand sync and find missing data and Apple Health: connect and manage your sources . Your data, your Claude freddy makes your own Apple Health data available to your own AI. You own it. You can pause a source, and if you delete your freddy account, your data is deleted. See Delete your account . FAQ Can Claude read my Apple Health data? Yes, through freddy. The freddy iPhone app reads Apple Health with the permissions you grant, and Claude reads that data from freddy once you add freddy from Claude's connector directory. Claude does not connect to Apple Health directly. Do I need to export a file from Apple Health? No. You install the freddy iPhone app and allow it to read Apple Health, so there is no export file to make or upload. The iPhone matters here: a desktop browser cannot read the Health database on your phone. Does this include my Apple Watch and other apps? Apple Watch data and records that other apps write to Apple Health can reach freddy, as long as you allow those categories. In freddy you can open the Apple Health source and see which apps and devices are contributing. Older records that never reached Apple Health cannot be recovered by a freddy re-import. What does it cost? Free is $0: 1 active source and 7 days of history shown, no card. Pro is $49 a year, and Believer is $249 once for life. Paid plans have no source limit and show your full history. You do not need a paid Claude plan. See pricing . Who owns the data, and can I delete it? You do. freddy makes your own data available to your own AI. If you delete your freddy account, your data is deleted. Steps are in Delete your account . Related Apple Health: connect and manage your sources Connect freddy to Claude Other apps through Apple Health or Health Connect Choose and connect a data source Understand sync and find missing data Apple Health MCP servers: 10 ways to get your Apple Health data into Claude, ChatGPT, or any AI client HRV across wearables: how to actually compare what your devices report Recipe: Why am I tired today? Sources freddy connects Pricing On Android? Health Connect Using ChatGPT instead? Connect freddy from ChatGPT freddy is an independent service and is not affiliated with, endorsed by, or sponsored by Apple or Claude.

## Health Connect: connect and manage your sources
URL: https://freddy.coach/help/connect-health-connect
Use freddy on Android to read supported records shared through Health Connect by your health and fitness apps.
Use freddy on Android to read supported records shared through Health Connect by your health and fitness apps. Before you connect Install freddy on the Android phone containing your Health Connect records. Use your personal phone profile. Health Connect is built into Android 14 and later; older compatible phones may need the Health Connect app. Connect Open freddy Sources after installing the phone app freddy for iPhone freddy for Android Open Health Connect on your phone and check that your recording app can write the categories you want. Open freddy → Sources → Health Connect and follow the permission prompts. Allow freddy to read the matching categories. Review history and background-access prompts if offered for your version and account. These are separate from category permissions. Keep freddy open for the first import, then inspect source names and dates in Data. What reaches freddy Supported activity, exercise, sleep, heart-rate and body records written into Health Connect. Supported nutrition and hydration records from apps that export them. The source app attached to each record, rather than every metric that app displays. History and updates Android normally limits an app’s initial historical reads to a recent window of about 30 days. Older access requires the history permission and support in your freddy version and account. Background access is separate and Android decides when tasks run. A longer query window does not itself grant phone permissions. If data is missing Your source app is connected but no data arrives: check its write permissions and actual entries in Health Connect. Only recent history appears: check the history permission and the dates actually stored in Health Connect. Sync works only when open: inspect background permissions and Android battery restrictions, then follow the Android background guide. Use it with your other sources Show my Health Connect activity next to my Withings measurements. Include the originating app for each activity record. Pause or disconnect Revoke freddy’s read permissions in Health Connect to stop phone access. Pausing a contributing source in freddy stops future import; Clear stored history is a separate action. Check its confirmation before deleting. Turning off permissions does not automatically erase the copies already stored in either app. Provider instructions Health Connect app permissions Android history and background reads Keep Android sync running Other bridge apps Related help Choose a source Find missing data and understand sync Contact freddy support

## Amazfit: connect, sync and troubleshoot
URL: https://freddy.coach/help/amazfit
Connect your Amazfit data through your Zepp account to compare workouts, sleep and daily activity with your other sources.
Connect your Amazfit data through your Zepp account to compare workouts, sleep and daily activity with your other sources. Before you connect Sync your watch in the Zepp app and use the Zepp account that holds those records. Connect Open Sources in freddy Open Amazfit in Sources and choose Connect. Sign in to the Zepp account containing your records. Review the permissions and authorize the categories you want to share. Return to freddy and inspect the source’s import status and date coverage. What reaches freddy Available daily activity and workout records from Zepp. Sleep and supported heart-rate, recovery or body measurements recorded by your devices. Additional measurements only when your device and the authorized Zepp service provide them. History and updates freddy imports the history made available by Zepp and receives new records through the connection. The date range can differ by category; device support determines which measurements exist. If data is missing A watch sync completed but data is absent: confirm the finished record is visible in the Zepp app before using Sync. Sleep appears but a different metric does not: compare that category in Zepp and check the permissions granted during connection. Use it with your other sources Compare my Amazfit sleep and activities with my Hevy training this week. Show dates and the source of each record. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Amazfit. You can also revoke freddy’s authorization in the source service’s connected-app settings. Provider instructions Amazfit and Zepp support Related help Choose a source Find missing data and understand sync Contact freddy support
Use your Amazfit data with your AI Pick your AI for the setup steps and ideas for what to ask. Claude On Claude Team and Enterprise, an Owner adds freddy for the organization first. ChatGPT OpenAI dots Availability depends on your OpenAI plan and region. Grok Bot Grok Muse Muse Code Gemini Gemini's custom apps need Gemini Spark, which comes with Google AI Pro or Ultra, and are US and English only. Gemini CLI Gemini CLI is a developer command-line tool, not the consumer Gemini app. The app connects separately and needs Gemini Spark. Claude Code Cursor Codex Perplexity Custom connectors require Perplexity Pro, Max, or Enterprise. OpenClaw Hermes Agent Instinct Hark the freddy CLI freddy is an independent service and is not affiliated with, endorsed by, or sponsored by the makers of the third-party AI apps above.

## Google Health: connect, sync and troubleshoot
URL: https://freddy.coach/help/google-health
Google Health is a direct account connection in alpha. It is separate from bringing Fitbit records through the Google Health mobile app and your phone’s health platform.
Google Health is a direct account connection in alpha. It is separate from bringing Fitbit records through the Google Health mobile app and your phone’s health platform. Before you connect Check the Google Health source card for alpha access. The presence of the source in the catalog does not mean the connection is enabled for your account. Connect Open Sources in freddy Open Google Health in freddy Sources and follow the access option shown. If Connect is enabled, sign in to the Google account holding your Google Health records and review the requested permissions. Return to freddy and check actual data arrival as well as the connection status. What reaches freddy Supported Google Health records made available to freddy after authorization. Available daily activity, sleep and health measurements for the enabled integration. Direct Google Health records remain attributable to that source, separate from Apple Health or Health Connect copies. History and updates Alpha availability, authorization and data arrival are separate steps. After access is enabled, the import can only read supported categories exposed for your account. Connecting this direct source does not grant freddy permission to read Health Connect on your phone. If data is missing Connect is unavailable or access is restricted: use the access option on the card. You connected the Google Health phone app to Health Connect but the direct source is empty: those are separate routes. Follow the Google Fitbit bridge guide for the mobile route. Both routes are connected: inspect their source labels and overlapping dates before combining totals. Use it with your other sources When direct Google Health data is available, compare its sleep records with my Hevy sessions. Keep phone-platform copies separate. Pause or disconnect Use Paused on the source to stop future sync without deleting its stored history. Disconnect removes the connection and its associated data from freddy; read the confirmation before continuing. It does not delete your records in Google Health. You can also revoke freddy’s authorization in the source service’s connected-app settings. Provider instructions Google Health developer platform Google Fitbit through the mobile app Related help Choose a source Find missing data and understand sync Contact freddy support

## Eight Sleep: connect through your phone
URL: https://freddy.coach/help/eight-sleep
Bring sleep records that Eight Sleep writes to Apple Health into freddy, then compare them with training and other sleep sources.
Bring sleep records that Eight Sleep writes to Apple Health into freddy, then compare them with training and other sleep sources. Before you connect This is a phone bridge, using Apple Health on iPhone. Install both Eight Sleep and freddy on the same phone and sign in to your own accounts. freddy does not sign in directly to Eight Sleep for this route. Connect the source app On iPhone In the Eight Sleep app, open its health integration settings and enable sharing to Apple Health. Review the write permissions for sleep in Apple Health. Finish the freddy Apple Health setup Then open freddy on that phone, connect the matching health platform and allow it to read the same categories. Keep freddy open while the initial import progresses. What reaches freddy Sleep records exported by the Eight Sleep app to Apple Health. Only the measurements present in those Health records; Pod settings, Autopilot controls and every in-app score are not part of this route. Eight Sleep continues to offer data export even though some integrations for importing other apps’ data have changed. Google Fit support does not establish a Health Connect route; this catalog connection uses iPhone. Check the connection After a night is processed in Eight Sleep, open Sleep in Apple Health and inspect a record’s source. It should identify Eight Sleep before you look for it in freddy. History and updates freddy reads supported records already written to the phone’s health platform. Eight Sleep decides which categories and dates it exports. Connecting today does not guarantee that its entire archive will be copied. Let the source app finish sharing before opening freddy; background updates depend on the phone’s scheduling. If data is missing Missing from the health platform: check Eight Sleep’s export setting, write permissions and the original record. freddy cannot import a record that has not reached the platform. Present on the phone but absent in freddy: check freddy’s read permissions, the source’s active state and import progress. Some fields are missing: compare the actual exported record with the original. Enabling a category does not create fields the app does not export. Use it with your other sources Compare my Eight Sleep records from Apple Health with my Garmin workout dates. Flag any nights missing from the export. Stop sharing or remove history Turn off Eight Sleep’s write permissions to stop new exports to the phone’s health platform, or pause its contributing source in freddy to stop future import there. Previously imported history is separate: use Clear stored history in freddy and read the confirmation for its scope. Revoking a permission alone does not delete the copies already stored. Provider instructions Eight Sleep data sharing Related help Choose a source Find missing data and understand sync Contact freddy support

## Google Fitbit: connect through your phone
URL: https://freddy.coach/help/google-fitbit
Use the Google Health mobile app to share Fitbit records with Apple Health or Health Connect, then let freddy read the records on your phone.
Use the Google Health mobile app to share Fitbit records with Apple Health or Health Connect, then let freddy read the records on your phone. Before you connect This is a phone bridge, using Apple Health on iPhone or Health Connect on Android. Install both Google Health and freddy on the same phone and sign in to your own accounts. freddy does not sign in directly to Google Fitbit for this route. Connect the source app On iPhone In the Google Health app, open Connections → Apps and services → Apple Health. Follow setup and allow the categories you want it to write. Finish the freddy Apple Health setup On Android In the Google Health app, open Connections → Partner apps and the option to sync health apps. Set up Health Connect and review the data-sharing permissions. Finish the freddy Health Connect setup Then open freddy on that phone, connect the matching health platform and allow it to read the same categories. Keep freddy open while the initial import progresses. What reaches freddy Supported activity, exercise, sleep, heart-rate and body records that the Google Health app writes to your phone’s health platform. Exported records only. A metric displayed in Google Health is not automatically available through both platforms. The source is named Google Fitbit in freddy; the companion phone app is now called Google Health. The separate Google Health source in freddy is a direct alpha integration, not a requirement for this bridge. Check the connection Sync your Fitbit device in the Google Health app. Check a finished sleep or activity entry in Apple Health or Health Connect and confirm its source before opening freddy. History and updates freddy reads supported records already written to the phone’s health platform. Google Fitbit decides which categories and dates it exports. Connecting today does not guarantee that its entire archive will be copied. Let the source app finish sharing before opening freddy; background updates depend on the phone’s scheduling. If data is missing Missing from the health platform: check Google Fitbit’s export setting, write permissions and the original record. freddy cannot import a record that has not reached the platform. Present on the phone but absent in freddy: check freddy’s read permissions, the source’s active state and import progress. Some fields are missing: compare the actual exported record with the original. Enabling a category does not create fields the app does not export. Use it with your other sources Compare Fitbit sleep shared through my phone with my Hevy workout dates. Name the originating app and the route. Stop sharing or remove history Turn off Google Fitbit’s write permissions to stop new exports to the phone’s health platform, or pause its contributing source in freddy to stop future import there. Previously imported history is separate: use Clear stored history in freddy and read the confirmation for its scope. Revoking a permission alone does not delete the copies already stored. Provider instructions Google Health with Apple Health Google Health with Health Connect Direct Google Health alpha Related help Choose a source Find missing data and understand sync Contact freddy support

## Cronometer: connect through your phone
URL: https://freddy.coach/help/cronometer
Bring exported Cronometer nutrition and measurements into freddy so food logs can sit next to your training and sleep.
Bring exported Cronometer nutrition and measurements into freddy so food logs can sit next to your training and sleep. Before you connect This is a phone bridge, using Apple Health on iPhone or Health Connect on Android. Install both Cronometer and freddy on the same phone and sign in to your own accounts. freddy does not sign in directly to Cronometer for this route. Connect the source app On iPhone In Cronometer, open More → Connect Apps & Devices → Apple Health. Review its write permissions for the categories you want to export. Finish the freddy Apple Health setup On Android In Cronometer, open More → Connect Apps & Devices → Health Connect. Enable the connection and grant write access for your chosen categories. Finish the freddy Health Connect setup Then open freddy on that phone, connect the matching health platform and allow it to read the same categories. Keep freddy open while the initial import progresses. What reaches freddy Nutrition and hydration records that Cronometer writes to the selected health platform. Supported biometrics you log and export, such as weight, where present. Exercise or sleep imported into Cronometer is not proof that Cronometer exports it back out. Cronometer’s Apple Health Backfill option imports Health data into the Cronometer diary. It is not a button for sending your entire food diary to freddy. Check the connection Choose a nutrition entry already saved in Cronometer and find the corresponding nutrition record in the health platform. Check the source and date, then look for the same coverage in freddy. History and updates freddy reads supported records already written to the phone’s health platform. Cronometer decides which categories and dates it exports. Connecting today does not guarantee that its entire archive will be copied. Let the source app finish sharing before opening freddy; background updates depend on the phone’s scheduling. If data is missing Missing from the health platform: check Cronometer’s export setting, write permissions and the original record. freddy cannot import a record that has not reached the platform. Present on the phone but absent in freddy: check freddy’s read permissions, the source’s active state and import progress. Some fields are missing: compare the actual exported record with the original. Enabling a category does not create fields the app does not export. Use it with your other sources Put my Cronometer daily nutrition beside my Garmin training time this week. Show logged days and missing days separately. Stop sharing or remove history Turn off Cronometer’s write permissions to stop new exports to the phone’s health platform, or pause its contributing source in freddy to stop future import there. Previously imported history is separate: use Clear stored history in freddy and read the confirmation for its scope. Revoking a permission alone does not delete the copies already stored. Provider instructions Cronometer and Apple Health Cronometer and Health Connect Related help Choose a source Find missing data and understand sync Contact freddy support

## MyFitnessPal: connect through your phone
URL: https://freddy.coach/help/myfitnesspal
Bring nutrition and weight records exported by MyFitnessPal into freddy through your phone’s health platform.
Bring nutrition and weight records exported by MyFitnessPal into freddy through your phone’s health platform. Before you connect This is a phone bridge, using Apple Health on iPhone or Health Connect on Android. Install both MyFitnessPal and freddy on the same phone and sign in to your own accounts. freddy does not sign in directly to MyFitnessPal for this route. Connect the source app On iPhone In MyFitnessPal, open More or your profile → Settings → Sharing & Privacy → HealthKit Sharing. Review the categories it can write. Later changes are made in Apple Health’s app permissions. Finish the freddy Apple Health setup On Android In MyFitnessPal, open Apps & Devices and choose Health Connect. Follow setup and allow the categories MyFitnessPal should write. Finish the freddy Health Connect setup Then open freddy on that phone, connect the matching health platform and allow it to read the same categories. Keep freddy open while the initial import progresses. What reaches freddy Shared food summaries and supported nutrients, plus weight where exported. On iPhone, MyFitnessPal does not send every nutrient or meal timestamp to Apple Health. Sleep and exercise displayed from Apple Health in MyFitnessPal are incoming data, not MyFitnessPal exports. On iPhone, food data is shared without the original meal timestamps; caffeine and vitamin A are excluded. Use daily comparisons rather than assuming a precise meal-time series. Check the connection Check a logged day in MyFitnessPal, then inspect the nutrition entry and its source in your health platform. A completed exercise import into MyFitnessPal does not establish this outgoing path. History and updates freddy reads supported records already written to the phone’s health platform. MyFitnessPal decides which categories and dates it exports. Connecting today does not guarantee that its entire archive will be copied. Let the source app finish sharing before opening freddy; background updates depend on the phone’s scheduling. If data is missing Missing from the health platform: check MyFitnessPal’s export setting, write permissions and the original record. freddy cannot import a record that has not reached the platform. Present on the phone but absent in freddy: check freddy’s read permissions, the source’s active state and import progress. Some fields are missing: compare the actual exported record with the original. Enabling a category does not create fields the app does not export. Use it with your other sources Compare my MyFitnessPal daily nutrition with my Garmin training by date. Do not invent meal times or fill unlogged days. Stop sharing or remove history Turn off MyFitnessPal’s write permissions to stop new exports to the phone’s health platform, or pause its contributing source in freddy to stop future import there. Previously imported history is separate: use Clear stored history in freddy and read the confirmation for its scope. Revoking a permission alone does not delete the copies already stored. Provider instructions MyFitnessPal Apple Health sharing MyFitnessPal Health Connect setup Related help Choose a source Find missing data and understand sync Contact freddy support

## Lose It!: connect through your phone
URL: https://freddy.coach/help/lose-it
Bring records Lose It! exports to your phone’s health platform into freddy, alongside sleep, activity and body measurements from other sources.
Bring records Lose It! exports to your phone’s health platform into freddy, alongside sleep, activity and body measurements from other sources. Before you connect This is a phone bridge, using Apple Health on iPhone or Health Connect on Android. Install both Lose It! and freddy on the same phone and sign in to your own accounts. freddy does not sign in directly to Lose It! for this route. Connect the source app On iPhone In Lose It!, open Profile → Automatic Tracking → Apple Health. Complete setup, then check Lose It!’s write permissions in Apple Health for the data you want to share. Finish the freddy Apple Health setup On Android In Lose It!, open Profile → Automatic Tracking → View All → Health Connect. Review Manage data and access and enable the appropriate write permissions for Lose It!. Finish the freddy Health Connect setup Then open freddy on that phone, connect the matching health platform and allow it to read the same categories. Keep freddy open while the initial import progresses. What reaches freddy Exported nutrition records and supported logged measurements. On Android, Lose It! supports writing nutrition and manually logged weight, exercise or steps when those permissions are enabled. Its calorie budget, bonuses and every diary detail are not guaranteed exported records. The items Lose It! is allowed to read control data coming into its diary. For this route to freddy, check the separate write permissions and actual exported entries. Check the connection Find an existing nutrition or weight entry from Lose It! in the health platform. Check its source and date before comparing it with freddy. History and updates freddy reads supported records already written to the phone’s health platform. Lose It! decides which categories and dates it exports. Connecting today does not guarantee that its entire archive will be copied. Let the source app finish sharing before opening freddy; background updates depend on the phone’s scheduling. If data is missing Missing from the health platform: check Lose It!’s export setting, write permissions and the original record. freddy cannot import a record that has not reached the platform. Present on the phone but absent in freddy: check freddy’s read permissions, the source’s active state and import progress. Some fields are missing: compare the actual exported record with the original. Enabling a category does not create fields the app does not export. Use it with your other sources Place my Lose It! daily food records next to my Withings measurements. Keep unlogged days and measurement dates explicit. Stop sharing or remove history Turn off Lose It!’s write permissions to stop new exports to the phone’s health platform, or pause its contributing source in freddy to stop future import there. Previously imported history is separate: use Clear stored history in freddy and read the confirmation for its scope. Revoking a permission alone does not delete the copies already stored. Provider instructions Lose It! Apple Health setup Lose It! Health Connect setup Related help Choose a source Find missing data and understand sync Contact freddy support

## Samsung Health: connect through your phone
URL: https://freddy.coach/help/samsung-health
Share Samsung Health records through Health Connect on Android, then import them with the freddy Android app.
Share Samsung Health records through Health Connect on Android, then import them with the freddy Android app. Before you connect This is a phone bridge, using Health Connect on Android. Install both Samsung Health and freddy on the same phone and sign in to your own accounts. freddy does not sign in directly to Samsung Health for this route. Connect the source app On Android In Samsung Health, open Settings and Health Connect. Review its connection and allow Samsung Health to write the data categories you want to share. Finish the freddy Health Connect setup Then open freddy on that phone, connect the matching health platform and allow it to read the same categories. Keep freddy open while the initial import progresses. What reaches freddy Supported activity, exercise, sleep and heart-rate records that Samsung Health writes to Health Connect. Other supported measurements only when they appear as exported Health Connect entries. Samsung Health Monitor features such as ECG are not implied by the standard Samsung Health connection. Sync a Galaxy Watch with Samsung Health first. The watch, Samsung Health app, Health Connect and freddy are separate steps in the path. Check the connection Open Health Connect and inspect a recent sleep or exercise record from Samsung Health. If the record is only visible on the watch, finish the Samsung Health sync first. History and updates freddy reads supported records already written to the phone’s health platform. Samsung Health decides which categories and dates it exports. Connecting today does not guarantee that its entire archive will be copied. Let the source app finish sharing before opening freddy; background updates depend on the phone’s scheduling. If data is missing Missing from the health platform: check Samsung Health’s export setting, write permissions and the original record. freddy cannot import a record that has not reached the platform. Present on the phone but absent in freddy: check freddy’s read permissions, the source’s active state and import progress. Some fields are missing: compare the actual exported record with the original. Enabling a category does not create fields the app does not export. Use it with your other sources Compare Samsung Health sleep shared through Health Connect with my Hevy workout dates. Show where each record came from. Stop sharing or remove history Turn off Samsung Health’s write permissions to stop new exports to the phone’s health platform, or pause its contributing source in freddy to stop future import there. Previously imported history is separate: use Clear stored history in freddy and read the confirmation for its scope. Revoking a permission alone does not delete the copies already stored. Provider instructions Samsung Health and Health Connect Related help Choose a source Find missing data and understand sync Contact freddy support

## Peloton: connect through your phone
URL: https://freddy.coach/help/peloton
Bring Peloton workouts shared with Apple Health or Health Connect into freddy to compare indoor sessions with your other training.
Bring Peloton workouts shared with Apple Health or Health Connect into freddy to compare indoor sessions with your other training. Before you connect This is a phone bridge, using Apple Health on iPhone or Health Connect on Android. Install both Peloton and freddy on the same phone and sign in to your own accounts. freddy does not sign in directly to Peloton for this route. Connect the source app On iPhone In the Peloton app, open its More or settings area and the Health app connection. Enable sharing, then review Peloton’s write permissions in Apple Health. Finish the freddy Apple Health setup On Android In the Peloton app, open the Health Connect integration when available and complete setup. In Health Connect, allow Peloton to write exercise and other categories you want to share. Finish the freddy Health Connect setup Then open freddy on that phone, connect the matching health platform and allow it to read the same categories. Keep freddy open while the initial import progresses. What reaches freddy Exported exercise sessions with the duration and other summary fields the app writes. Active energy and heart rate when included in the shared records. Leaderboard positions, class detail and full power traces are not guaranteed through this route. Use the Peloton account containing the completed session. Support for a health platform is not a promise that every class type or every field is exported. Check the connection After the session is saved in Peloton, inspect its exercise entry in the health platform. Check duration and source before searching for it in freddy. History and updates freddy reads supported records already written to the phone’s health platform. Peloton decides which categories and dates it exports. Connecting today does not guarantee that its entire archive will be copied. Let the source app finish sharing before opening freddy; background updates depend on the phone’s scheduling. If data is missing Missing from the health platform: check Peloton’s export setting, write permissions and the original record. freddy cannot import a record that has not reached the platform. Present on the phone but absent in freddy: check freddy’s read permissions, the source’s active state and import progress. Some fields are missing: compare the actual exported record with the original. Enabling a category does not create fields the app does not export. Use it with your other sources Compare my Peloton sessions with my Garmin outdoor rides this week. Identify overlapping recordings before totaling time. Stop sharing or remove history Turn off Peloton’s write permissions to stop new exports to the phone’s health platform, or pause its contributing source in freddy to stop future import there. Previously imported history is separate: use Clear stored history in freddy and read the confirmation for its scope. Revoking a permission alone does not delete the copies already stored. Provider instructions Peloton health-platform integration Peloton on Health Connect Related help Choose a source Find missing data and understand sync Contact freddy support

## Lifesum: connect through your phone
URL: https://freddy.coach/help/lifesum
Bring nutrition and hydration records shared by Lifesum into freddy to compare logged days with your training and sleep.
Bring nutrition and hydration records shared by Lifesum into freddy to compare logged days with your training and sleep. Before you connect This is a phone bridge, using Apple Health on iPhone or Health Connect on Android. Install both Lifesum and freddy on the same phone and sign in to your own accounts. freddy does not sign in directly to Lifesum for this route. Connect the source app On iPhone In Lifesum, open Settings → Automatic Tracking or Automatic Measurement and activate Apple Health. Review the write permissions. Water export also has an option under Settings → Water Settings. Finish the freddy Apple Health setup On Android In Lifesum, open Diary or Progress → Profile → Settings → Automatic Tracking → Health Connect. Allow it to write the nutrition and hydration categories you want. Finish the freddy Health Connect setup Then open freddy on that phone, connect the matching health platform and allow it to read the same categories. Keep freddy open while the initial import progresses. What reaches freddy Nutrition and water records written by Lifesum to the selected health platform. Only the exported records; Lifesum’s calorie target and in-app guidance are not the same as logged intake. Exercise calories and steps read into Lifesum do not establish an outgoing export of those categories. The sync direction matters: Lifesum uses incoming exercise to update its own view. For freddy, verify outgoing nutrition or water entries in your health platform. Lifesum currently excludes food logged with Multimodal (AI) Tracking from its Health Connect export. Check Customize Diary if those entries are missing. Check the connection Find a saved Lifesum nutrition or water record in Apple Health or Health Connect, including its source and date, then compare that date in freddy. History and updates freddy reads supported records already written to the phone’s health platform. Lifesum decides which categories and dates it exports. Connecting today does not guarantee that its entire archive will be copied. Let the source app finish sharing before opening freddy; background updates depend on the phone’s scheduling. If data is missing Missing from the health platform: check Lifesum’s export setting, write permissions and the original record. freddy cannot import a record that has not reached the platform. Present on the phone but absent in freddy: check freddy’s read permissions, the source’s active state and import progress. Some fields are missing: compare the actual exported record with the original. Enabling a category does not create fields the app does not export. Use it with your other sources Show my Lifesum logged nutrition days next to my Oura sleep. Keep gaps visible and do not treat targets as intake. Stop sharing or remove history Turn off Lifesum’s write permissions to stop new exports to the phone’s health platform, or pause its contributing source in freddy to stop future import there. Previously imported history is separate: use Clear stored history in freddy and read the confirmation for its scope. Revoking a permission alone does not delete the copies already stored. Provider instructions Lifesum Health Connect setup Lifesum Apple Health setup Related help Choose a source Find missing data and understand sync Contact freddy support

## Sleep Cycle: connect through your phone
URL: https://freddy.coach/help/sleep-cycle
Share Sleep Cycle sleep records to Apple Health on iPhone, then use freddy to compare those nights with your other sources.
Share Sleep Cycle sleep records to Apple Health on iPhone, then use freddy to compare those nights with your other sources. Before you connect This is a phone bridge, using Apple Health on iPhone. Install both Sleep Cycle and freddy on the same phone and sign in to your own accounts. freddy does not sign in directly to Sleep Cycle for this route. Connect the source app On iPhone In Sleep Cycle, open Profile → More → Apple Health → Connect. Review the permission to write sleep data in Apple Health. Finish the freddy Apple Health setup Then open freddy on that phone, connect the matching health platform and allow it to read the same categories. Keep freddy open while the initial import progresses. What reaches freddy Sleep records that Sleep Cycle writes to Apple Health. Available fields in the exported sleep record; audio recordings, alarm settings and every Sleep Cycle score are not included by this connection. Sleep Cycle records its own sleep. Its Apple Health integration does not fill a night you forgot to record by importing another app’s sleep. Apple Health integration is listed among Sleep Cycle’s free features. Check the connection After a tracked night is saved, inspect Sleep in Apple Health and confirm an entry from Sleep Cycle. Then open freddy and inspect the same date. History and updates freddy reads supported records already written to the phone’s health platform. Sleep Cycle decides which categories and dates it exports. Connecting today does not guarantee that its entire archive will be copied. Let the source app finish sharing before opening freddy; background updates depend on the phone’s scheduling. If data is missing Missing from the health platform: check Sleep Cycle’s export setting, write permissions and the original record. freddy cannot import a record that has not reached the platform. Present on the phone but absent in freddy: check freddy’s read permissions, the source’s active state and import progress. Some fields are missing: compare the actual exported record with the original. Enabling a category does not create fields the app does not export. Use it with your other sources Compare my Sleep Cycle nights from Apple Health with my Garmin activity dates. Flag nights with no exported record. Stop sharing or remove history Turn off Sleep Cycle’s write permissions to stop new exports to the phone’s health platform, or pause its contributing source in freddy to stop future import there. Previously imported history is separate: use Clear stored history in freddy and read the confirmation for its scope. Revoking a permission alone does not delete the copies already stored. Provider instructions Sleep Cycle Apple Health connection Sleep Cycle feature availability Related help Choose a source Find missing data and understand sync Contact freddy support

## Renpho: connect through your phone
URL: https://freddy.coach/help/renpho
Share measurements from the Renpho Health app through Apple Health or Health Connect, then bring them alongside your other sources in freddy.
Share measurements from the Renpho Health app through Apple Health or Health Connect, then bring them alongside your other sources in freddy. Before you connect This is a phone bridge, using Apple Health on iPhone or Health Connect on Android. Install both Renpho and freddy on the same phone and sign in to your own accounts. freddy does not sign in directly to Renpho for this route. Connect the source app On iPhone Open the Renpho Health app’s connected-app settings and enable Apple Health. In Apple Health’s app permissions, allow Renpho Health to write the measurement categories you want. Finish the freddy Apple Health setup On Android In the Renpho Health app, open the connected-app or health-sharing settings and select Health Connect when available. Grant write access for your chosen measurements. Finish the freddy Health Connect setup Then open freddy on that phone, connect the matching health platform and allow it to read the same categories. Keep freddy open while the initial import progresses. What reaches freddy Supported measurements such as weight and body fat when written by Renpho Health. Only categories exported to the health platform. A scale’s full set of body scores may remain in the Renpho app. Use the Renpho Health app and the person profile assigned to the measurement. An older Renpho app’s Google Fit connection is not evidence that Health Connect has received the record. Check the connection Take an existing scale reading that is visible under your profile in Renpho Health and find it in the phone’s health platform. Compare date, unit and source before looking in freddy. History and updates freddy reads supported records already written to the phone’s health platform. Renpho decides which categories and dates it exports. Connecting today does not guarantee that its entire archive will be copied. Let the source app finish sharing before opening freddy; background updates depend on the phone’s scheduling. If data is missing Missing from the health platform: check Renpho’s export setting, write permissions and the original record. freddy cannot import a record that has not reached the platform. Present on the phone but absent in freddy: check freddy’s read permissions, the source’s active state and import progress. Some fields are missing: compare the actual exported record with the original. Enabling a category does not create fields the app does not export. Use it with your other sources Show Renpho weight measurements beside my Garmin weekly activity. Keep each measurement date and unit attached. Stop sharing or remove history Turn off Renpho’s write permissions to stop new exports to the phone’s health platform, or pause its contributing source in freddy to stop future import there. Previously imported history is separate: use Clear stored history in freddy and read the confirmation for its scope. Revoking a permission alone does not delete the copies already stored. Provider instructions Renpho Health platform support Renpho Health help Related help Choose a source Find missing data and understand sync Contact freddy support

## Pull your data with the freddy Personal API (beta)
URL: https://freddy.coach/help/personal-api
A read-only REST API to pull your own freddy health data into your own tools. Currently in beta.
What it is freddy's Personal API is a read-only REST API for pulling your own health data into your own tools: a Grafana dashboard, a Google Sheet, a script, or a personal app you built. It is the code-facing companion to your MCP connection, which is for talking to your data through an AI. Everything the API returns is your own already-connected data, bounded by your plan's history window. Beta access The Personal API is in beta. The “API keys” card appears in freddy under AI apps automatically, no request needed. Learn more at freddy.coach/api. Get a key Open AI apps in freddy. In the API keys card, enter a label (for example grafana) and create a key. Copy the key. It starts with fk_live_ and is shown only once. Store it in your tool's secret store, never in code you share. You can create up to 10 keys and revoke any of them at any time. Authenticate Send the key as a bearer token on every request: curl -H "Authorization: Bearer fk_live_..." https://freddy.coach/api/v1/profile Endpoints GET /api/v1/metrics: your normalized health metrics, newest first. GET /api/v1/catalog: the distinct metrics, providers, and units available to you. GET /api/v1/sources: your connected sources. GET /api/v1/profile: your profile and plan window. The full reference, with every parameter, is at https://freddy.coach/api-docs, and the machine-readable spec is at https://freddy.coach/api/openapi.json (import it into Postman, Insomnia, or your own tooling). Read your metrics curl -H "Authorization: Bearer fk_live_..." \ "https://freddy.coach/api/v1/metrics?from=2026-06-01&limit=100" Each row is a small object: date, timestamp, provider, metric, device, deviceLabel, sourceId, numValue, textValue, unit. timestamp is the exact moment in UTC, and date is its day. device is the source device as freddy stores it, and deviceLabel is the friendly name you see in freddy. sourceId is the source's own id for the record, so rows with the same provider and sourceId belong to the same workout or night. Narrow the results with: metric: one or more metric names, repeated or comma-separated, for example metric=sleep_score,hrv. provider: one or more providers, case-insensitive, for example provider=garmin,oura. device: one or more stored device values (the device field, not the label). It needs metric, for example metric=steps&device=Forerunner 965. provider is optional. source_id: one or more record ids from sourceId. It needs provider, for example provider=garmin&source_id=12345. from and to: ISO dates. A from earlier than your plan window is clamped to the window rather than rejected. limit: page size, up to 1000 (default 200). The meta block Every response includes a meta block describing the plan window that produced it: "meta": { "plan": "FREE", "history_window_days": 7, "history_floor": "2026-07-02", "source_cap": 1 } On a paid plan the window fields are null, meaning unbounded. This tells your tool exactly how far back the data goes, so nothing is silently truncated. Read everything (pagination) Responses include next_cursor. To pull a full range, keep calling with the cursor until it comes back null: FK="fk_live_..." cursor="" while true; do page=$(curl -s -H "Authorization: Bearer $FK" \ "https://freddy.coach/api/v1/metrics?limit=1000${cursor:+&cursor=$cursor}") echo "$page" | jq -c '.data[]' cursor=$(echo "$page" | jq -r '.next_cursor') [ "$cursor" = "null" ] && break done Raw records Add include_raw=true to get the full record behind each row: exercise sets, sleep stages, heart-rate streams, whatever the source sent. It's the same data your AI reads through freddy, at full resolution. curl -H "Authorization: Bearer fk_live_..." \ "https://freddy.coach/api/v1/metrics?metric=workout_raw&include_raw=true" Each row then has raw and rawStatus: ok: raw holds the record exactly as the source sent it. none: this row has no record attached. Another row of the same workout often has it. unavailable: there is a record, but we couldn't read it. Try again later; if it stays unavailable, that record can't be read. A few things that help: One workout or night is usually several rows. Group them by provider and sourceId. Where the record sits depends on the source. Garmin, Withings and Intervals.icu attach it to one row of the record, and Oura to every row. For Apple Health and Health Connect, ask for the metric that ends in _raw, for example workout_raw. Keep following next_cursor to get every page. Rate limits If you receive a 429, wait and retry. Sync into your own tools Because the API is plain HTTP with a bearer key, most tools can read it directly. A common setup is a scheduled Google Apps Script that pulls /api/v1/metrics into a sheet each morning, or a Grafana or personal database that queries it on a timer. If you would like a ready-made Google Sheets script to start from, ask us in the chat. Good to know Read-only. The API never changes your data or your connections. Retired sources are not returned. Revoking a key stops it working immediately, with no caching.

## Choose and connect a data source
URL: https://freddy.coach/help/choose-a-source
Find the setup guide for every freddy source, including direct connections, iPhone and Android bridges, and alpha access.
Start with the source that already holds your records. A direct connection reads the source account; a phone bridge reads the records an app shares with Apple Health or Health Connect. Connect an account Authorize a source account or paste its API credentials in freddy. Each guide covers setup, available data, history and troubleshooting. Polar Oura Withings Intervals.icu Suunto Concept2 Wahoo Ultrahuman Hevy · API credentials Lyfta · API credentials Runalyze · API credentials Airthings · API credentials WHOOP Garmin Connect Amazfit · beta Connect your phone’s health platform Use the freddy app on the phone that contains your records. The browser can display uploaded data; phone access is granted on the device. Apple Health · iPhone Health Connect · Android Connect through another app These sources reach freddy through your phone’s health platform. The manual explains the app’s export step and the platform permission step. Eight Sleep · iPhone Google Fitbit · iPhone and Android Cronometer · iPhone and Android MyFitnessPal · iPhone and Android Lose It! · iPhone and Android Samsung Health · Android Peloton · iPhone and Android Lifesum · iPhone and Android Sleep Cycle · iPhone Renpho · iPhone and Android Other bridge apps: check compatibility and set up sharing Connections in alpha These manuals explain restricted access and the setup path when enabled. A catalog listing does not mean the source is available to connect. Dexcom · alpha Google Health · alpha After your first connection Check that records have arrived, including their dates and source names. Connect another source you want to compare with the first. Connect your AI client, ask a dated question and request source labels. Connect an AI client Understand sync and find missing data Import Garmin archive and FIT data Keep Android sync running Which route should you use? For a source with a direct connection, use its manual to check the available data and history. For apps that export through your phone, use the bridge route. If you use both, inspect overlapping records before asking for totals. Different sources can describe the same workout or night. Open Sources

## Understand sync and find missing data
URL: https://freddy.coach/help/sync-and-missing-data
Trace a missing record, check import progress and permissions, and distinguish stored history from an AI query window.
Start with one missing record and follow its path. A connection status shows access; the import status and dates show what has arrived. 1. Check the original record Open the app or service that recorded it. Confirm it is saved, assigned to your account and fully synced from the device. Note the date, category and source without sharing private health values in a public report. 2. Check the route Direct connection: open its source in freddy and inspect authorization, Active or Paused, and the import progress. Phone bridge: first find the exported record in Apple Health or Health Connect. Then check freddy’s read permission for that category. Archive import: distinguish the file upload from the processing that follows. Keep the upload page open until it confirms upload completion. 3. Check freddy Data Look for the source and date in Data. An imported record can use a different series from a live record. A provider may supply daily summaries while its own app displays a detailed chart. Open Data Sync or re-import? Sync checks for updates through that source’s supported route. Start here when recent records are missing. Re-import all history requests the history currently available through the connection. It cannot extend a provider’s archive window, recreate records never exported to a phone, or grant missing permissions. A running import may take longer for a large archive or a provider that limits requests. Check progress before starting another one. History has more than one limit The source controls what it shares. Your phone controls what an app may read. freddy’s import determines what has arrived. Your account’s query window determines what your AI client can request. A shorter AI query window does not mean freddy deleted older stored data. If the AI answer is missing records Ask for the exact date range, source names and available records before requesting a comparison. If the data is visible in freddy, check that your AI client is connected to the same freddy account and that the dates are in its allowed query window. AI connection troubleshooting If numbers do not match Check units, timezone, daily boundaries and the original data type. A sleep score is not sleep duration; a training estimate is not a measured value. If two sources recorded the same session, check overlap before adding totals. Before disconnecting Use Paused to stop future sync when you want to keep history. Disconnect can delete the source’s stored data in freddy. It is not the first step for a slow import. Phone permission revocation and Clear stored history are separate actions; read the confirmation for the scope. Contact support Send the source name, phone platform if relevant, missing category, approximate date and the status or error text you see. Keep API keys, tokens and private measurements out of the message. Open freddy support Choose a source manual

## Changelog: One clear state for every source (2026-10-30)
Each source now shows one clear state, Listening, Syncing, Paused or Action needed, on the web and in the iPhone app. Overview shows all your sources as logos with a status dot, and Activity lists only the sources with something going on. On iPhone, Apple Health workouts with several parts, such as a triathlon, now keep each part. You can now change your login email on the web, in Settings. freddy sends a code to your current inbox and one to your new inbox, and your data, sources and AI connections stay as they are. The app gets it in its next update. The Garmin history import on the web now brings in past workouts of every sport, such as strength, yoga, walking and hiking, from your Garmin data export. The app gets it in its next update.

## Changelog: Your direct URL no longer contains your key (2026-10-23)
Your direct URL no longer has your key in it. It's tied to your freddy account, and an app that can't sign in sends your MCP key as a header instead. Your old direct URL keeps working for now.

## Changelog: Full records on the Personal API (2026-10-14)
The Personal API can now return the full record behind each metric, like exercise sets, sleep stages and heart-rate streams, along with the device and record id for every row. freddy now knows your time zone on the web too, and asks before it changes a zone you chose.

## Changelog: freddy is an official Claude connector (2026-10-11)
freddy is now an official Claude connector. Add freddy from Claude's connector directory, connect your freddy account, and use your health and performance data in any Claude conversation.

## Changelog: Lyfta and Runalyze are open to everyone (2026-09-25)
Lyfta and Runalyze are out of beta, so anyone can connect them from Sources. Runalyze shares data through its Personal API, which needs a Runalyze Supporter or Premium account.

## Changelog: Choose the account when you reconnect an AI app (2026-09-24)
When you reconnect an AI app, freddy now shows which account you're signed in to, and you can switch to another one.

## Changelog: A focused flow for adding a source (2026-09-18)
Adding a source on the dashboard now runs as its own flow with only the steps that source needs, and setup you did not finish is waiting for you when you come back. Overview stops showing the setup complete notice once you are done with it.

## Changelog: Hevy history imports are more reliable (2026-09-17)
Hevy history now comes across from your saved workout list, including older sessions that could be missed before. Edits and deletions you make in Hevy afterwards are matched up too, so what freddy holds keeps agreeing with your Hevy log.

## Changelog: Garmin summaries keep every activity type (2026-09-16)
When Garmin reported two activity types for the same stretch of time, one could replace the other. freddy now keeps the detail from both. This applies to activity summaries arriving from now on.

## Changelog: The dashboard speaks eight more languages (2026-09-15)
Account settings and navigation now come in German, Spanish, French, Italian, Japanese, Korean, Polish and Simplified Chinese, alongside English. Pick a language in Settings, or let freddy follow your browser. Source setup, data views, support, emails and help articles stay in English for now.

## Changelog: Step by step AI setup, and a searchable help center (2026-09-14)
Connecting an AI client now goes one step at a time on the dashboard, for every client freddy works with. The help center has search and topic navigation, with direct starting points for connecting a source, connecting an AI client, or tracking down data that is missing.

## Changelog: Amazfit opens up, and Google Health arrives in early access (2026-09-11)
Amazfit is now open to Pro and Believer accounts with nothing to request. Connect your Zepp account from Sources on the dashboard. Google Health can now connect to freddy directly. Access is limited, and you can join the waitlist from Sources. Source cards on the dashboard have been redesigned, so connection status, sync progress and the everyday actions take less searching for.

## Changelog: More Oura data, and Runalyze history picks up where it stopped (2026-09-06)
Oura now brings across stress, resilience, cardiovascular age, pulse wave velocity, VO2 max, tags, ring details and battery history, alongside the sleep, readiness, activity, workout, heart rate and blood oxygen data already there. An older Oura connection keeps syncing what it already has access to. If a newer part is missing, Sources asks you to reconnect, and freddy then fills in that history. A large Runalyze history sync now continues where it stopped instead of starting over.

## Changelog: Steadier Apple Health syncing in the latest app (2026-09-05)
Apple Health syncing is more reliable, and a history import reports its progress more clearly. Update the freddy app from the App Store to get this, along with a batch of stability fixes across the app.

## Changelog: Sources show when your data actually landed (2026-09-04)
A source card now shows two things instead of one: when the last sync finished, and when data actually arrived. A sync can succeed on a day your source has nothing new to hand over. A history import that is running now shows real progress instead of a status label that never moves. Connecting Garmin now walks you through its optional history permission, which Garmin leaves unticked, before its authorization screen opens.

## Changelog: Amazfit and Zepp, in early access (2026-08-28)
freddy can now connect to Amazfit through your Zepp account and bring across sleep, activity and workouts. Access is limited. Join the waitlist from Sources on the dashboard to be in the first group.

## Changelog: Gemini and Cursor connect, and a Garmin import fix (2026-08-24)
The Gemini app can now connect to freddy. Adding freddy needs Gemini Spark on a paid Google AI plan and a computer to set it up on, and Google keeps custom apps to the US and English for now. Cursor can connect too. It could not finish adding freddy at all before, and that is fixed. Uploading a Garmin account export could fail at the upload step, with a message blaming the file when the file was fine. Fixed for everyone. Nothing to redo on your side, just upload again.

## Changelog: Apple Health sleep and workouts stay current (2026-08-20)
Apple Health sleep and workouts could stop updating while everything else kept arriving. The latest iOS update reworks how they catch up, and they now recover on their own. During a large history import, today's numbers now land first instead of waiting for the backlog to finish. Both fixes arrive with the latest iOS update, out now on the App Store. The Android update follows shortly.

## Changelog: See how a Suunto workout unfolded (2026-08-18)
Suunto workouts now carry your pace, heart rate, cadence, power and elevation through the session, not just the totals. You can ask where you slowed down, or how your heart rate responded to a climb. Your existing Suunto workouts fill in on their own over the next few days.

## Changelog: Suunto workouts now include your route (2026-08-13)
Your Suunto workouts now bring across the route you actually ran or rode, not just the summary. You can ask your AI assistant which runs followed the same route, and how your pace and climbs compared. Your existing Suunto workouts fill in automatically, including older ones.

## Changelog: freddy tells you when a source stops syncing (2026-08-10)
The freddy app can now send you a notification when one of your connected sources goes quiet, so a source that quietly stopped sending data does not sit unnoticed until you next look. The alert names the source that stopped, so you know where to look when your other sources are still coming through fine. Alerts arrive at a sensible hour where you are, and you can switch them off any time in Settings on the app or the dashboard.

## Changelog: Garmin running dynamics (2026-08-07)
Ground contact time, vertical oscillation, vertical ratio, ground contact balance, and stride length now flow into freddy from your Garmin runs. Ask Claude how your form held up across a run, or how it's trending over time. Cycling and swim dynamics from Garmin come through too.

## Changelog: Fitbit now syncs on iPhone, and Hevy body measurements (2026-08-06)
Fitbit data now reaches freddy on iPhone. Turn on Apple Health sharing in the Google Health app, then connect Apple Health in freddy, and steps, sleep, heart rate and workouts come through automatically. Heart rate variability is not part of what Google shares today. Hevy now brings across the body measurements you log in the app: weight, body fat, lean mass, and measurements like waist, chest and arms. Past entries come across too, so you can ask how your measurements have moved alongside your training.

## Changelog: Garmin history imports are more resilient (2026-08-03)
A hiccup part way through a Garmin history import no longer throws away the whole import. The rest of your files keep going. If an import does stop early, we now tell you what was saved instead of only saying it failed.

## Changelog: Android sync fixes and battery guidance (2026-07-29)
Fixed Health Connect background sync getting stuck when another app wrote an invalid record. Sync now skips it and keeps going. The Android app now detects battery settings that block background sync and shows how to fix them for your phone brand. Still importing your history now shows on Android while your Health Connect history import is running.

## Changelog: Apple Health sync fix (2026-07-26)
Fixed Apple Health syncing stalling for part of your metrics while a few kept updating. Update the iOS app and affected accounts recover automatically over a few days of normal use. Garmin ZIP import now handles archives exported on a Mac and reports partially imported files clearly.

## Changelog: One-click install, and the Personal API opens in beta (2026-07-22)
Add freddy to Claude in one click. The new Install in Claude button opens Claude with freddy prefilled, ready to authorize. You will find it on the dashboard, the homepage, and in the app. Install in ChatGPT takes you straight to freddy in the ChatGPT app directory. Pasting your link by hand still works everywhere. The Personal API is now in beta, included in Pro. Create a key in the dashboard and pull your sleep, workouts, glucose and recovery into your own dashboards, spreadsheets and scripts. A new page at freddy.coach/api shows what you can build, with a quickstart and the full API reference.

## Changelog: freddy now recognizes the apps behind your Apple Health and Health Connect data (2026-07-20)
Apps that send data through Apple Health or Health Connect, like Garmin Connect, AutoSleep, Yazio, MacroFactor and many more, now show up by name in freddy with their own tile. Sync helper apps are labeled as such, so you always know which app actually measured your data. Source labels on your existing data got cleaner too. Nothing about the data itself changed.

## Changelog: Apple Health accuracy (2026-07-15)
Apple Health metrics now read more accurately, including blood oxygen percentages and daily activity totals.

## Changelog: See the apps behind your Apple Health and Health Connect data (2026-07-14)
Apps that sync through Apple Health or Health Connect, like Cronometer, MyFitnessPal, and Peloton, now show up as their own sources in freddy, each with its logo and latest status. Open Apple Health or Health Connect to see every app feeding it, and clear any app's saved history whenever you want.

## Changelog: Daily glucose and vitals trends from Apple Health and Health Connect (2026-07-10)
Glucose from a CGM connected through Apple Health or Health Connect now shows daily averages, highs, lows, and time in range, so you can ask how it's trending across weeks or months instead of scrolling through raw readings. Spot readings like blood oxygen, blood pressure, and body temperature get the same daily summaries. Health data from Apple Health and Health Connect syncs more reliably. A metric can no longer get stuck and stop updating. Recent glucose and workouts now appear faster.

## Changelog: WHOOP is now available to everyone (2026-07-09)
Connect WHOOP and freddy can read your recovery, strain, sleep, and workouts in any conversation. Your recovery and strain now sit right next to everything else you track, so your AI can reason across all of it in one place.

## Changelog: freddy is now on iPhone (2026-07-08)
Install freddy from the App Store and bring your iPhone health data into any AI you use. Connect Apple Health to pull in sleep, heart rate, workouts, and recovery from your Apple Watch and the apps that write to Health. Background sync keeps your data fresh without reopening the app.

## Changelog: freddy is now on Android (2026-07-03)
Install freddy from the Google Play Store and bring your Android health data into any AI you use. Connect Health Connect to pull in steps, heart rate, sleep, and workouts from Samsung Health, Google Fit, and Fitbit. Background sync keeps your data fresh without reopening the app.

## Changelog: freddy is now a ChatGPT app (2026-06-26)
freddy is now an official app in the ChatGPT app store. Find it in ChatGPT's app directory and connect in a couple of clicks. It works in ChatGPT on web and mobile. Once connected, just ask about your sleep, training and recovery inside any chat.

## Changelog: Interval breakdowns from intervals.icu (2026-06-25)
freddy can now read the breakdown of each interval in your structured intervals.icu workouts, including power, heart rate and pace per rep, not just the session totals.

## Changelog: Ultrahuman is now available to everyone (2026-06-23)
Anyone can now connect their Ultrahuman Ring AIR or M1 to AI. Your ring's sleep stages, sleep score, and blood oxygen now flow into your data.

## Changelog: Beta integrations now come with Pro (2026-06-22)
Pro and Believer plans now include access to new integrations while they're in beta. Connect beta sources as soon as they land, no waitlist.

## Changelog: Richer Polar workout data (2026-06-17)
freddy now reads more from your Polar training sessions, including training load, your GPS route, and other workout detail. Your Polar fitness profile, like VO2max and training thresholds, is now available to your AI assistant too.

## Changelog: A clearer home screen, and richer workout detail (2026-06-15)
Your home now opens to live stats and trends for your connected sources and the AI tools using freddy, instead of plain lists. Add a source or connect an AI right from there. Your AI can now see how your heart rate, pace, and power moved through a workout, not just the totals. Ask things like how your heart rate held up on a long ride, or where your pace fell off in the back half. Available for Garmin and Polar workouts.

## Changelog: Import your full Garmin history (2026-06-11)
Garmin users can now import their complete history. Request a copy of your data from Garmin, upload it on the Connections page, and years of sleep, training, and wellness data become part of your freddy history. Imported history merges cleanly with your live sync, so your AI can reason over long term trends, not just recent weeks. If one of your connections ever needs reconnecting, freddy now flags it on your dashboard with a one click reconnect that keeps all your synced history.

## Changelog: Garmin is here (2026-06-04)
Connect your Garmin account and freddy reads your sleep, training, heart rate, stress, and recovery, right inside any AI chat. New Garmin data syncs automatically as your device updates.

## Changelog: freddy on the command line (2026-06-02)
You can now use freddy straight from your terminal. Sign in once, connect your wearables, and pull your health data with a single command. It is built for AI agents too. Point your assistant at freddy and it can keep your data flowing and act as your coach or trainer, with no setup work on your side.

## Changelog: Your full Polar history (2026-06-01)
Polar now pulls years of your past data, not just what arrives after you connect. Open your dashboard and reconnect Polar to import your full sleep, training, recovery, and heart rate history.

## Changelog: More stable wearable connections (2026-05-31)
Fixed a bug that could disconnect a wearable and ask you to reconnect more often than it should.

## Changelog: Wahoo and Concept2 are now generally available (2026-05-28)
Wahoo (ELEMNT, TICKR, KICKR, RIVAL) is out of beta and open to all accounts. No waitlist needed. Concept2 Logbook (rowing, SkiErg, BikeErg) is out of beta and open to all accounts. Freddy has a new look. New logo across the app, emails, and website. Introducing BELIEVER, a one-time lifetime purchase that unlocks every PRO feature with no annual renewal.

## Changelog: New integration: Ultrahuman (private beta) (2026-05-27)
Connect your Ultrahuman Ring AIR or M1 CGM to bring your recovery, HRV, glucose, and metabolic data into Freddy. Available in private beta. Reach out if you'd like early access.

## Changelog: Data from multiple devices is no longer overwritten (2026-05-22)
If you have more than one device on the same account, such as a watch and a sleep monitor, each night's data from every device is now kept separately instead of one replacing another. You can now ask about a specific device by name.

## Changelog: Intervals.icu activities now use the correct timestamp (2026-05-20)
Fixed Intervals.icu activities showing at the wrong time. New activities are stored correctly from now on. To clean up older ones, disconnect and reconnect the integration.

## Changelog: Withings sleep data fixes (2026-05-15)
Withings sleep now reports its full deep, light, and REM breakdown, matching what you see in the Withings app. Body temperature from ScanWatch 2 now syncs.

## Changelog: Garmin, Wahoo, and Concept2 in beta, plus easier sign-in (2026-05-13)
Garmin (beta) — sleep, HRV, stress, body battery, activity, and recovery, with women's health data if enabled in Garmin Connect. Wahoo Cloud (beta) — ELEMNT, TICKR, KICKR, and RIVAL activities sync automatically. Concept2 Logbook (beta) — rowing, SkiErg, and BikeErg workouts. Sign-in now uses a 6-digit code instead of a magic link, so it works when the email opens in a different browser than the one you're signing into. Claude can now ask about specific date ranges in your history — useful for "how was my training in Q1 2024" or year-over-year comparisons. Freddy now pauses inside Claude if our Terms have changed since you signed up. Open the dashboard and accept to continue.

## Changelog: Faster data on connect, and Claude knows when a sync is in progress (2026-05-10)
When you connect a new wearable, your recent data is now usable within minutes. Full history backfills in the background — no waiting. Claude now tells you when data is still syncing instead of returning "no data found" mid-backfill.

## Changelog: Sign in to Freddy from headless agents (2026-05-08)
Agents like OpenClaw, Claude Code, and Cursor can now sign in to Freddy via a device code — your agent shows a short code, you paste it into Freddy once, and the agent is connected to your wearable data. Same flow works for any MCP client that runs outside a browser.

## Changelog: New help center (2026-05-07)
Freddy now has a help center at freddy.coach/help with setup guides for each wearable and AI client.

## Changelog: Suunto out of beta, plus Oura profile data (2026-05-06)
Suunto is now available to everyone, no waitlist. Oura profile data (age, weight, height, sex) is now queryable by your AI, so it can answer questions like "what's my max heart rate based on age?"

## Changelog: In-app support chat, Notion AI, and a sync polling fix (2026-05-04)
New "Help" tab to message us from inside Freddy. Replies come back in-app and via email. Notion AI Custom Agents can now connect to Freddy (Business and Enterprise plans). Sync progress now keeps updating after switching browser tabs.

## Changelog: OAuth sign-in and guided AI setup (2026-05-03)
ChatGPT and any other MCP client that supports OAuth can now sign in to Freddy with a standard authorize screen, no token pasting. New "Connected apps" panel shows every app you've authorized with a one-click Revoke. The dashboard now opens with a 3-step setup checklist and tabbed guides for ChatGPT, Claude, Perplexity, and other clients.

## Changelog: Recent activity panel (2026-04-30)
New "Recent activity" panel on the dashboard shows recent calls to your MCP URL — when, which tool, and where the request came from. Helps you spot unauthorized use. Rotate your MCP token from Settings if anything looks off. Full audit log available as a CSV export.

## Changelog: Hevy out of beta (2026-04-28)
Hevy is now available to everyone, no waitlist.

## Changelog: Suunto (beta) and email preferences (2026-04-27)
Suunto integration in beta: workouts, sleep, recovery, and 24/7 health data. Request access on the Suunto tile. New "Email me about Freddy" toggle in Settings to control product update emails. Transactional emails always send.

## Changelog: Email alerts when a connection breaks (2026-04-16)
If your wearable's authorization expires or is revoked, you'll get an email with a one-click reconnect link. No more silent gaps in your data.

## Changelog: Hevy integration (beta) (2026-04-15)
Connect Hevy to bring strength workouts into Freddy, with full per-set detail. Available in beta. Click "Request access" on the Hevy tile to join the waitlist.

## Changelog: Browse beta integrations and request new ones (2026-04-14)
Beta integrations are now visible to everyone in the connections list. Click "Request access" to join the waitlist. New "Missing something?" row lets you suggest any wearable, scale, or health app you'd like Freddy to support.

## Changelog: Intervals.icu integration (2026-04-11)
Connect Intervals.icu to bring training analytics, wellness data, and activity history into Freddy. New rides, runs, and uploads sync in automatically.

## Changelog: Oura Ring out of beta (2026-04-10)
Oura Ring is now available to everyone, no waitlist.

## Changelog: More wearables, and the PRO plan (2026-04-09)
Oura, Withings, WHOOP, and Dexcom now supported (WHOOP and Dexcom in beta). Upgrade to PRO for unlimited connections and full data history. Dashboard and website now mobile-friendly.

## Changelog: Freddy is live (2026-03-25)
Connect your Polar watch and get a personal MCP URL for Claude. Ask Claude about your sleep, workouts, recovery, and training data in any conversation. Passwordless sign-in. Dashboard to manage your connection.

## See which meetings spike your stress
URL: https://freddy.coach/recipes/meeting-stress
Line up your intraday stress with your calendar to find which meetings, and which people, wind you up.
Needs: Intraday stress (one of: garmin)
Also needs (same chat): Google Calendar
See what stress data you have
First, let your assistant discover which stress metrics are on your account and how far back they go. This also confirms your watch actually records the all-day stress signal.
Run list_metrics and show me which stress and body-battery metrics I have, with their date ranges and source.
Always run list_metrics first, because metric names differ per device.
Pull one busy day at full detail
Ask for a single meeting-heavy weekday at 3-minute resolution. The raw payload carries the intraday stress curve, not just the daily average.
Using query_metrics with include_raw, pull stress_durationInSeconds for last Tuesday (if that day's raw payload is not there, pull stress_avgStressLevel instead). Read the timeOffsetStressLevelValues series. Ignore any negative values (-1 to -5 mean no reading, not calm). Convert each point to my local wall-clock time using startTimeInSeconds and startTimeOffsetInSeconds, and describe how my stress moved through the day.
Line it up with your calendar
Now bring in your calendar connector in the same conversation and let the assistant overlay your meetings on the stress curve.
Pull my Google Calendar events for that same day. For each meeting, tell me my average stress during it (excluding no-reading values) and how it compares to my daily baseline. Which meetings, and which attendees, line up with my biggest stress climbs?
You'll get a readout roughly like this: a calm morning baseline, a sharp climb through a back-to-back block of afternoon meetings, and a slow recovery afterwards. Meetings with certain attendees or certain formats (large group, status reviews) tend to show up as the steepest climbs. The numbers in this recipe are illustrative, not from a real day.
Garmin only today. Garmin is the one source freddy holds a true intraday stress signal for. WHOOP cannot do this, because its API exposes no intraday stress or heart-rate data at all. Your calendar connector must be active in the same conversation as freddy, or the assistant only sees the stress curve with no meeting labels. Negative stress values (-1 to -5) are gaps (off-wrist, motion, not enough data), not low stress. Averaging them in will quietly skew the result. Attribute the data as Garmin.

## Why am I tired today?
URL: https://freddy.coach/recipes/morning-readiness
A one-prompt morning briefing that reads last night's sleep, HRV and resting heart rate and tells you how hard to push today, whatever ring or watch you wear.
Needs: Sleep and recovery (one of: oura, garmin, whoop, polar, withings, ultrahuman, apple-health, health-connect)
Ask for your morning readout
One prompt does it. Your assistant reads whatever recovery signals your device records overnight and turns them into a plain-language readout, with no jargon and no dashboard.
Run list_metrics, then read my most recent night: sleep duration and stages, resting heart rate, HRV, and any readiness or recovery score I have. Compare each to my own 30-day baseline, not to population norms. In plain language, tell me whether today looks like a day to push hard, keep it easy, or rest, and which single signal is driving that.
Always run list_metrics first, because these metrics are named differently on every device.
Ask why a number is off
If a signal looks off, ask the assistant to trace it back through the week.
My HRV or resting heart rate is off my baseline this morning. Look back over the last week of my data and tell me what most likely explains it: a late night, alcohol, a hard workout, illness, or travel. Use only what is in my data, and say so if there is not enough to tell.
You'll get something like: slept a little short, resting heart rate a touch high, HRV below your baseline, so today reads as an easy day, with the short sleep as the main driver. The specific numbers depend on your night. The example here is illustrative, not from a real day.
HRV is not comparable across brands. Different devices use different HRV measures, so compare your HRV to your own history, never to someone else's number or another device. Apple Health and Health Connect give sleep, HRV and resting heart rate but no single readiness or recovery score. If you connect only those, the briefing works from the raw signals instead of a score. The readout is only as good as what your device records. A night the ring or watch missed shows as gaps, not zeros.

## Does your bedroom air wreck your sleep?
URL: https://freddy.coach/recipes/air-quality-sleep
Line up your bedroom CO2, humidity and temperature with how well you slept, night after night, to find the environmental thing quietly costing you sleep.
Needs: Air quality (one of: airthings)
Needs: Sleep (one of: oura, garmin, whoop, polar, withings, ultrahuman, apple-health, health-connect)
See your overnight air and your sleep
Your air monitor records the bedroom through the night; your ring or watch records how you slept. First, confirm both are flowing and pick a stretch of nights to look at.
Run list_metrics. Confirm I have bedroom air readings (CO2, humidity, temperature, and anything else like VOCs or particulates) and a nightly sleep score or sleep stages. Show me the date range where they overlap.
Run list_metrics first; air-sensor and sleep metric names differ by device.
Correlate air with sleep quality
Now let the assistant line up each night's overnight air against that night's sleep.
For the last 30 nights, take my overnight bedroom readings (roughly the hours I was asleep) and compare them to that night's sleep score, deep sleep, and number of awakenings. Which environmental factor lines up most with my worse nights: high CO2, high or low temperature, high humidity, or something else? Rank them and tell me how strong each link looks.
Test a change
Found a likely culprit? Change one thing, then check back in a week or two.
For the next 10 nights I am going to change one thing in my bedroom (for example crack a window, lower the thermostat, or run a dehumidifier). Afterwards, compare those nights to the 10 before on the same air and sleep measures, and tell me whether it actually helped.
A typical result: your worst sleep scores cluster on the nights your bedroom CO2 climbed highest, with temperature a secondary factor and humidity showing little link. Your own home will differ. The example is illustrative, not from a real night.
This is correlation, not proof. A link between stuffy air and poor sleep is a strong hint, not certainty. The test-a-change step is what turns a hunch into something you can trust. You need enough nights. A week is rarely enough to see a pattern; three to four weeks is better. Align the overnight window. Whole-day air averages hide the hours that matter, so ask the assistant to use roughly your sleep hours, not the 24-hour average.

## See what spikes your glucose
URL: https://freddy.coach/recipes/glucose-response
Find out how your glucose actually responds to specific meals, workouts or your morning coffee, using your own readings instead of generic advice.
Needs: Glucose (one of: dexcom, ultrahuman, apple-health, health-connect)
Find your glucose data
Whether you wear a continuous monitor or log the occasional finger-stick, your assistant can work with what you have. Start by seeing what is there.
Run list_metrics and show me my glucose data: the metric names, the units (mg/dL or mmol/L), how often it is recorded, and the date range.
Units differ by source: Apple Health stores mg/dL, Health Connect stores mmol/L. Tell the assistant to keep them straight.
Look at a specific response
Pick something you want to understand, a meal, a walk, a coffee, and see what your glucose did around it.
On [date], I [ate X / went for a walk / had coffee] at roughly [time]. Pull my glucose around that window and describe what happened: the starting level, how high it went, how long it took to come back down, and whether it dipped below baseline afterwards.
Compare choices
Once you have a few of these logged, let the assistant compare them.
Compare my glucose response to [oatmeal] versus [eggs] across the times I have logged each. Which one raises my glucose more, and which one keeps it steadier? Use only my readings, and say so if I do not have enough logged to compare.
A typical read: a bowl of oats sends you up sharply and back down with a dip an hour later, while eggs barely move the line, so eggs keep you steadier. Your body will respond in its own way. The example is illustrative, not from real readings.
Units differ by source. Apple Health uses mg/dL and Health Connect uses mmol/L. If your data comes from more than one, make sure the assistant does not mix them. Continuous vs occasional. A continuous monitor shows the full curve; occasional finger-stick readings only show snapshots, so the more you log around an event, the clearer the picture. One response is not a rule. Sleep, stress, movement and time of day all shift how you respond to the same food, so look for patterns across several logs, not a single day.

## High-carb vs low-carb: how food moves your recovery
URL: https://freddy.coach/recipes/carbs-recovery
See whether your higher-carb days or your lower-carb days leave you better recovered and better rested the next morning, using your own logged food.
Needs: Nutrition (one of: apple-health, health-connect)
Needs: Recovery or sleep (one of: oura, garmin, whoop, polar, withings, ultrahuman)
Get your food into freddy
freddy reads the daily nutrition totals your food-logging app writes to Apple Health or Health Connect. Log a couple of weeks of meals in whichever app you like (for example Cronometer or MyFitnessPal) with it syncing to Apple Health or Health Connect, then confirm the numbers are flowing.
Run list_metrics and confirm I have daily nutrition data: carbohydrates, protein, fat and calories. Show me how many days I have logged and the date range.
freddy stores nutrition as daily totals, not per meal, so log through an app that syncs to Apple Health or Health Connect.
Split high-carb from low-carb days
Let the assistant sort your logged days by how much carbohydrate you ate.
Using my logged days, split them into my higher-carb third and my lower-carb third by grams of carbohydrate. For each group, show my average next-morning recovery: HRV, resting heart rate and sleep score. Do higher-carb days leave me better or worse recovered the next morning?
Control for the obvious
Carbs travel with other things. Ask the assistant to check whether it is really the carbs.
Before I trust that, check whether my high-carb days were also my hard-training days or my late-eating days, since those would muddy the result. If you can, compare high vs low carb only within similar training days.
You might find your higher-carb evenings line up with slightly better sleep and steadier HRV, or the opposite. Either way you get your own answer instead of a generic rule. The direction is personal. The example is illustrative, not from your real logs.
Nutrition is daily totals only. freddy sees the day's carbs, protein, fat and calories, not individual meals or their timing. For meal-level questions you need a continuous glucose monitor or a food-timing log. You have to log food. This only works if you record meals in an app that syncs to Apple Health or Health Connect, and a few scattered days will not be enough. Aim for two weeks or more. Correlation, not causation. Carbs ride along with training load, eating time and alcohol, so the control step is what keeps you from blaming the wrong thing.

## A triathlon plan that adjusts itself
URL: https://freddy.coach/recipes/adaptive-triathlon-plan
Hold your training plan in a project, then check it every morning against what your body actually did and change today's session when plan and reality disagree.
Needs: Training sessions (one of: garmin, polar, whoop, suunto, wahoo, concept2)
Needs: Sleep and recovery (one of: oura, garmin, whoop, polar, withings, ultrahuman, apple-health, health-connect)
Also needs (same chat): Google Calendar
See what your account actually has
A plan is only as good as the data behind it. Start by letting your assistant discover which session, sleep and recovery metrics you have and how far each one goes back, so the plan is built on your real training rather than on what you tell it you have been doing.
Run list_metrics and show me which workout, sleep, recovery and training load metrics I have, with their date ranges and source.
Metric names differ per source, so always run this first.
Give the project your athlete profile
Create a project and put the durable facts in its instructions: race date and distance, the weekly hours you can realistically train, which days are immovable, injury history, your weakest discipline, and how blunt you want it to be. Long-lived facts only. This week's session does not belong here, it belongs in the plan file and on your calendar.
Build the block from your real training
Ask for the plan inside the project, with freddy switched on, and make it justify the volume against what you have actually been doing rather than a textbook ramp. Keep the plan as a file in the project so every later run starts from the same document instead of re-deriving it.
Using freddy, look at my actual training over the last 8 weeks, then build a periodised plan through to my race with a repeating weekly structure. Justify the starting volume against what I have really been doing, not a standard ramp, and tell me where you think the plan is most likely to break. Save it as a plan file in this project.
Put the week on your calendar
Have it write the coming week's sessions to your calendar with the session type, target duration and intensity in the event body. Write actions are off by default. Until they are enabled for your calendar connector, the assistant can read your week but cannot put anything on it or move anything.
Create this week's sessions from the plan file on my calendar. Put the session type, target duration and intensity in the event body so I can read it from my phone without opening the plan.
Run the morning check
This is the loop. Schedule it for after you are up and your sources have synced, not before. Last night's sleep does not exist anywhere until your watch or ring syncs, so an early-hours run reports on the night before last. Let it run for a week before you trust it, and read what it changed rather than skimming the summary. Expect to tune the project instructions early: the usual failure is turning cautious after one poor night, which one line fixes, telling it a single bad night is noise and to downgrade only on a trend or after a genuinely hard day.
Using freddy, check whether I completed yesterday's planned session: pull yesterday's workouts with type, duration and intensity and compare them against what the plan called for. Then pull last night's sleep and this morning's recovery, including HRV and resting heart rate, plus my training load across the last 7 days, compared against my own 30 day baseline rather than population norms. Then look at today's planned session on my calendar and decide whether it still makes sense. If it does, confirm it in one line. If it does not, change the calendar event to the session I should actually do, put the reason in the event body, and update the plan file. Tell me what you changed and why, naming which source each conclusion came from. If two sources disagree, say so rather than averaging them. Never compare HRV across different sources. If anything is genuinely ambiguous, such as an unrecorded session or a workout matching no planned one, ask me instead of guessing.
Schedule this as a recurring run. Keep it narrow and save longer reasoning for a weekly review.
A short daily readout: whether yesterday's session happened as planned, partly, or not at all; where your recovery sits against your own recent baseline; and one committed call for today rather than a hedge. When it changes something, you also get the calendar event rewritten, with the reason in the event body, and the plan file updated to match. The useful surprise is how often it catches sessions you quietly did not do. It is easy to tell yourself you are on plan, harder when something reads yesterday's file and notices the intervals came in short. Any figures you see in examples of this recipe are illustrative, not readings from a real account.
The calendar write-back needs an assistant that can both run on a schedule and edit calendar events. Elsewhere the same prompts still work as a read-and-propose loop: you get the verdict and the suggested change, and you make the edit. Write actions are off by default for connectors. Until they are enabled, the assistant can read your calendar but not change it. Your calendar connector must be active in the same conversation as freddy, or the assistant sees your training data with nothing to compare it against. Schedule the check for after you wake and your sources sync. Last night's sleep does not exist anywhere until the watch or ring uploads it, so a very early run reports on the night before last. Never let it compare HRV across sources. Garmin, WHOOP, Ultrahuman and Health Connect report RMSSD while Apple Health reports SDNN, and these are different measurements rather than different units of the same thing, so a chart mixing them looks meaningful while being nonsense. Training load and readiness feeds can cover a shorter or staler window than your raw activities, so the assistant may have yesterday's session without yesterday's load. Tell it to say so rather than filling the gap. Attribute the data as Garmin when it comes from a Garmin device. This is not a coach and not a clinician. It is good at holding a plan and noticing when reality has drifted from it. Do not use it to train through pain or to rehabilitate an injury.

## A daily symptom scan
URL: https://freddy.coach/recipes/daily-symptom-scan
Spot early signals you're run down by comparing resting heart rate, HRV, sleep, temperature and respiratory rate against your own baseline.
Needs: Sleep and recovery (one of: oura, garmin, whoop, polar, withings, ultrahuman, apple-health, health-connect)
Run the daily scan
One prompt does it. freddy compares last night and the past three days with your 30-day baseline across the signals that often move first when you are coming down with something or carrying too much load, then tells you whether today is a day to take it easy.
Use freddy to compare last night and the past three days with my 30-day baseline. Flag changes in resting heart rate, HRV, sleep, temperature and respiratory rate, note any missing data, and tell me whether to take it easy today.
Works best after your ring or watch has synced last night's data.
You'll get something like: resting heart rate a little high for the third morning, HRV below your baseline, sleep a touch short, skin temperature and respiratory rate still near your norm. The short sleep and elevated resting heart rate are the main flags, so today reads as an easy day. Gaps are called out rather than filled in. The example here is illustrative, not from a real day.
Not every source records temperature or respiratory rate. If a signal is missing, the scan should say so rather than inventing a value. HRV is not comparable across brands. Different devices use different HRV measures, so compare your HRV to your own history on the same source, never to another device. Schedule or run this after your sources have synced. Last night's sleep does not exist anywhere until the ring or watch uploads it, so a very early run reports on the night before last. This is not a clinician. It is good at noticing when your own signals drift from your baseline. Do not use it to diagnose illness.