Intervals.icu MCP Server
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Intervals.icu MCP Servershow my recent activities"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Intervals.icu MCP Server
A self-hosted, multi-tenant Model Context Protocol (MCP) server that exposes the
Intervals.icu API — activities, events, wellness, power curves, gear, and
custom items — as MCP tools for Claude. It runs as a remote, OAuth-authenticated connector at
https://intervalsicu.farhoodlabs.com/mcp, where each user's Intervals.icu API key and athlete id are
stored encrypted and looked up per request from their OAuth identity.
This is a fork of the upstream single-tenant project (mvilanova/intervals-mcp-server), reworked for remote multi-user operation.
How it fits together
The system is three services on one Kubernetes host (intervalsicu.farhoodlabs.com), deployed via Flux
GitOps and sharing one CloudNativePG Postgres database:
Service | Path | What it does |
intervalsicu-mcp (this repo) |
| The MCP server. Verifies OAuth JWTs and serves Intervals.icu data as MCP tools over streamable-HTTP. |
intervalsicu-mcp-ui |
| Portal where users sign in (Google via Better Auth) and set their Intervals.icu credentials; admins approve accounts. |
intervalsicu-mcp-auth |
| Better Auth OAuth 2.1 server: dynamic client registration (DCR), PKCE, EdDSA-signed JWTs + JWKS, Google login. |
Connection flow: the Claude connector self-registers at the auth server (DCR), the user logs in with
Google, and Claude receives a JWT access token whose audience is the /mcp URL. This server verifies
that JWT against the auth server's JWKS and resolves the caller's stored Intervals.icu credentials by the
token subject. A user who hasn't been approved or hasn't set credentials gets a friendly "account isn't
ready" message.
Related MCP server: fitness-mcp-server
Connecting Claude
Add a custom remote connector in Claude pointing at:
https://intervalsicu.farhoodlabs.com/mcpName it without a dot — use intervals, not intervals.icu. Claude Desktop and mobile fail to load
the tools if the connector name contains a dot. Claude handles registration and Google login for you.
Your account must be approved in the portal and have Intervals.icu credentials set.
Available tools
get_activities, get_activity_details, get_activity_intervals, get_activity_streams,
get_activity_messages, add_activity_message, get_events, get_event_by_id, add_or_update_event,
delete_event, delete_events_by_date_range, get_wellness_data, get_athlete_power_curves,
get_gear_list, get_custom_items, get_custom_item_by_id, create_custom_item, update_custom_item,
delete_custom_item.
Local development
Requires Python 3.12+ and uv.
uv venv --python 3.12 && source .venv/bin/activate
uv sync --all-extras # install with dev extras
uv run pytest # tests (enforces a >=90% coverage gate)
ruff check . # lint
mypy src tests # type checkRun locally over stdio (single-user, using the env fallback credentials below):
python -m intervals_mcp_server.serverFor local HTTP without OAuth, set MCP_TRANSPORT=http and leave MCP_ISSUER / MCP_RESOURCE /
MCP_JWKS_URI unset — auth is disabled when those aren't all present, and tools use the API_KEY /
ATHLETE_ID env fallback.
Environment variables
Local / single-user fallback (used only when there is no OAuth context):
API_KEY— an Intervals.icu API key (Settings → API on intervals.icu).ATHLETE_ID— your athlete id, e.g.i12345(from the intervals.icu URL). Must matchi?\d+.INTERVALS_API_BASE_URL— optional, defaults tohttps://intervals.icu/api/v1.
Multi-tenant / deployment:
DATABASE_URL— async Postgres DSN (postgresql+asyncpg://…) for the shared user store.INTERVALS_ENC_KEY— base64-encoded 32-byte AES-256-GCM key used to encrypt stored API keys. Shared with the UI service so both read/write the same ciphertext.MCP_ISSUER,MCP_RESOURCE,MCP_JWKS_URI— OAuth: token issuer, the/mcpresource URL, and the JWKS endpoint (an internal in-cluster URL, to bypass Cloudflare). Setting all three enables auth.MCP_TRANSPORT—stdio(default),sse, orhttp/streamable-http. Deployment useshttp.MCP_STATELESS_HTTP,MCP_JSON_RESPONSE— HTTP tuning. The deployment sets bothtrue(stateless + JSON) so the distributed connector can hit any replica; stateful/SSE breaks it.FASTMCP_HOST,FASTMCP_PORT,FASTMCP_LOG_LEVEL— HTTP bind + logging.
Deployment
Pushes to main trigger Gitea CI (.gitea/workflows/build.yaml): it runs the test suite with the
coverage gate, then builds and pushes a container image to
git.farh.net/farhoodlabs/intervalsicu-mcp. Flux reconciles the new image onto the cluster. Database
migrations run as an Alembic (alembic upgrade head) initContainer before the server starts.
License
GNU General Public License v3.0 (inherited from the upstream project).
This server cannot be deployed
Maintenance
Related MCP Connectors
tri-train is a training platform for triathletes that syncs activities from Strava and Garmin and exposes them to Claude through a remote MCP server. Ask Claude how your last block went, why your run pace dropped, or whether you are fresh enough for Sunday's race, and it answers from your real data: activities with laps and grade-adjusted pace, weekly load and the performance management chart (CTL/ATL/TSB), sleep, HRV and resting heart rate, and your race calendar with predictions. Claude can also act: create and adjust planned workouts with structured intervals, push them to your Garmin watch, add races and write notes. Every change is journaled and can be undone. Read and write access are separate OAuth scopes, and the server enforces them on every call. Built and hosted in Switzerland by a triathlete; data stays on our own servers (Swiss revDSG / GDPR).
Connect Claude to your Intervals.icu watch data for fitness, workout review, and plan writing.
Multi-tenant hosted MCP server for Oura Ring — 21 read-only tools, OAuth per user.
Pace is a remote MCP server that exposes wearable and fitness data to Claude via the Model Context Protocol. It connects to Garmin, Oura, Whoop, Polar, Fitbit and 20+ devices and provides 15 tools for querying sleep, activity, recovery, and training data. Hosted on Google Cloud Run, OAuth 2.1 authentication, Streamable HTTP transport. Instructions: First you need to create an account at: https://pacetraining.co and connect your wearables. After that you can connect the remote Server via Custom Connector in Claude and OAuth 2.1 Flow startet.
Related MCP Servers
- AlicenseAqualityFmaintenanceA Model Context Protocol (MCP) server for Intervals.icu integration. Access your training data, wellness metrics, and performance analysis through Claude and other LLMs.4837MIT
- FlicenseNot gradedqualityDmaintenanceProvides MCP servers for Claude to access fitness data from Strava and intervals.icu, enabling natural language queries for activity analysis, advanced training metrics, and wellness tracking.1-
- AlicenseAqualityCmaintenanceMCP server for Intervals.icu that enables AI assistants to manage athletic training data, including activities, calendar events, wellness metrics, and workout libraries.19Apache 2.0
- AlicenseAqualityDmaintenanceAn MCP server that lets Claude plan cycling training in intervals.icu, upload workouts to the calendar, and review planned versus actual outcomes, fitness trends, and power curves.9MIT