Self-hosted sync and dashboards for your Strava activities: PostgreSQL storage, training-load metrics (TRIMP, CTL, ATL, TSB, FTP estimates), and Grafana visualization.
TrainingPulse is not affiliated with, endorsed by, or sponsored by Strava. “Strava” is a trademark of Strava, Inc. Each person who runs this stack must register their own Strava API application and comply with the Strava API Agreement and Strava API Brand Guidelines.
This project is largely AI-generated. The numbers it computes — including training load, fitness, fatigue, form, and FTP estimates — are informational only and are not medical advice. Verify anything you plan to act on, and consult a qualified professional for health or training decisions.
Example views from the bundled dashboards.
 Activity detail |
 Fitness & Freshness |
 Account overview |
- Docker and Docker Compose
- A Strava API application: create one at strava.com/settings/api and note the Client ID and Client Secret. Set the Authorization Callback Domain to the hostname where you will run the stack (for example
localhostormyserver.local).
-
Configure your environment. Copy the example file and fill in your Strava credentials, a strong database password, and the URL the app will be reachable at:
cp .env.example .env
At minimum, set
STRAVA_CLIENT_ID,STRAVA_CLIENT_SECRET,POSTGRES_PASSWORD, andAPP_BASE_URL(used to build the OAuth callback). -
Start the stack. From the repo root:
docker compose up -d
This brings up four containers: the FastAPI app on port
8000, Grafana on port3000, the MCP service on port8001, and PostgreSQL (internal only). -
Connect Strava. Open the app at
http://your-host:8000/and click the official Connect with Strava button to complete OAuth. The initial backfill starts automatically — the same page shows live progress, and ongoing syncs run every 15 minutes from then on. -
Open Grafana. Browse to
http://your-host:3000/. The Postgres data source (strava-pg) and the bundled dashboards are provisioned automatically, so the dashboards appear without any manual setup. -
Optional: connect an MCP client. The read-only MCP service is available at
http://your-host:8001/mcpfor Cursor or another MCP-aware client. It exposes training summaries, activity lookup, gear usage, and sync-health tools over your network; keep it on a trusted network because tool results can include private training data. If you connect through a homeserver hostname or LAN IP, add it toMCP_ALLOWED_HOSTSin.env.
Pulled from Strava and stored locally:
- Activity summaries (name, sport, times, distance, elevation, average/max HR and power, kilojoules, calories, device and gear, suffer score) — merged from the list and detail endpoints.
- Raw telemetry streams (heart rate, power, cadence, temperature, grade, GPS, etc.) for every activity that has them.
- OAuth tokens (auto-refreshed) and your Strava athlete profile values used for HR zones and FTP, unless overridden in the environment.
Computed by this project:
- Per-activity TRIMP (heart-rate zone-weighted training load), with a sport-based estimate when an activity has no HR data.
- Time spent in each heart-rate and power zone per activity.
- Best 20-minute power and an estimated FTP (95% of best 20-minute power).
- Daily CTL (Fitness), ATL (Fatigue), and TSB (Form)
Out of the box Max HR is 190, Resting HR is 60, and FTP comes from your Strava profile (falling back to 200). Override any of them by setting MAX_HR, REST_HR, or FTP in your .env. See DEVELOPERS.md for the full precedence rules, including how to let Strava's profile values drive Max HR and zones.
For architecture, the HTTP API, the full list of environment variables, and how each metric is calculated, see DEVELOPERS.md.
This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License v3.0 as published by the Free Software Foundation. See the LICENSE file for the full text.