Wahoo Fitness Cloud API — fetch workouts, download FIT files, parse power/HR/cadence/GPS into local SQLite for analysis
Install with the open skills CLI (global, non-interactive — available in every Claude Code session):
npx skills add LeoYeAI/openclaw-master-skills --skill "wahoo-cloud" -g -a claude-code -yOr manually — clone and copy the skill directory (SKILL.md + companion files):
git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills /tmp/openclaw-master-skills && cp -r /tmp/openclaw-master-skills/skills/wahoo-cloud ~/.claude/skills/wahoo-cloudThis skill is a directory: SKILL.md is the entry point; the files below ship with it.
---
name: wahoo-cloud
description: Wahoo Fitness Cloud API — fetch workouts, download FIT files, parse power/HR/cadence/GPS into local SQLite for analysis
homepage: https://cloud-api.wahooligan.com/
metadata: {"clawdbot":{"emoji":"🚴","requires":{"bins":["python3"],"env":["WAHOO_CLIENT_ID","WAHOO_CLIENT_SECRET"]},"primaryEnv":"WAHOO_CLIENT_ID"}}
---
# Wahoo Cloud API Skill
Programmatic access to the Wahoo Fitness Cloud API for ELEMNT BOLT/ROAM/ACE head units. Fetches workout metadata, downloads FIT files from Wahoo's CDN, and parses ride data (power, cadence, HR, GPS, elevation) into a local SQLite database.
API base: `https://api.wahooligan.com`. Workout endpoints live under `/v1/workouts`. OAuth2 with the `offline_data` scope yields a long-lived refresh token; access tokens expire after ~2 hours and the skill auto-refreshes on 401.
## Agent quickstart (read this first)
If you're an agent invoking this skill on behalf of a user:
| User asks | Run this |
|---|---|
| "Sync my Wahoo workouts" / "Pull new rides" | `python3 {baseDir}/scripts/fetch_workouts.py` |
| "Show recent rides" / "Last week's training" | Query `~/.openclaw/workspace/training/wahoo.db` (or `$WAHOO_TRAINING_DIR/wahoo.db`) — schema below |
| "Parse this FIT file" | `python3 {baseDir}/scripts/parse_fit.py PATH.fit [--summary-only]` |
| "Set up Wahoo" / "Connect my Wahoo" | Walk the user through Setup §1–3 below; then run `python3 {baseDir}/scripts/oauth_setup.py` |
| "Refresh my Wahoo token" | `bash {baseDir}/scripts/refresh_token.sh` (only needed if auto-refresh fails) |
The fetch script is **idempotent** — safe to run on a heartbeat. It skips workouts already fully synced (`fit_parsed_at IS NOT NULL`). Sandbox rate limits (25 req / 5 min) trigger automatic backoff, so a first sync of a long history may take many minutes.
The skill **cannot** ship credentials. Each user needs their own Wahoo developer app — no shortcut. Setup is a one-time browser handshake.
**Credential auto-loading:** if `WAHOO_CLIENT_ID` / `WAHOO_CLIENT_SECRET` aren't in the calling shell, `wahoo_auth.py` automatically reads them from `~/.openclaw/secrets/wahoo.env` (override path with `$WAHOO_ENV_FILE`). This means an OpenClaw agent can invoke `fetch_workouts.py` without sourcing anything — token refresh just works.
## Setup
### 1. Register a Wahoo Developer App
1. Go to https://developers.wahooligan.com
2. Create an application (Sandbox is automatic — no review)
3. Set callback URL (e.g. `https://localhost:8080/` — the manual-paste OAuth helper works with any registered callback)
4. Request scopes: `workouts_read offline_data user_read` (add `power_zones_read plans_read routes_read` if you want zones/plans/routes)
5. Note your **Client ID** and **Client Secret**
### 2. Configure Credentials
Add to `~/.clawdbot/clawdbot.json`:
```json
{
"skills": {
"entries": {
"wahoo": {
"enabled": true,
"env": {
"WAHOO_CLIENT_ID": "your-client-id",
"WAHOO_CLIENT_SECRET": "your-client-secret",
"WAHOO_REDIRECT_URI": "https://localhost:8080/"
}
}
}
}
}
```
Or as environment variables:
```bash
export WAHOO_CLIENT_ID="..."
export WAHOO_CLIENT_SECRET="..."
export WAHOO_REDIRECT_URI="https://localhost:8080/"
```
### 3. Run OAuth2 Flow
```bash
python3 {baseDir}/scripts/oauth_setup.py
```
The script prints an authorization URL. Open it in a browser, log in with your Wahoo account, approve. You'll be redirected to your callback URL with `?code=...` in the query string (the page itself will fail to load — that's expected; just copy the URL or the `code` value). Paste it back into the script. It exchanges the code for tokens and writes them to `~/.openclaw/secrets/wahoo_tokens.json`.
### 4. Fetch Workouts
```bash
python3 ~/.openclaw/workspace/training/fetch_wahoo.py
```
This pulls the workout list, fetches detail (and FIT URL) for each new workout, downloads FIT files into `~/.openclaw/workspace/training/wahoo_fit/`, parses them, and upserts records into `~/.openclaw/workspace/training/wahoo.db`.
## Usage
### List Workouts (paginated)
```bash
curl -s -H "Authorization: Bearer ${WAHOO_ACCESS_TOKEN}" \
"https://api.wahooligan.com/v1/workouts?page=1&per_page=30"
```
Response shape: `{ workouts: [...], total, page, per_page, order, sort }`. Note that `workout_summary` is `null` in the list response — fetch detail per workout to get summary + FIT URL.
### Get Workout Detail (with FIT URL)
```bash
curl -s -H "Authorization: Bearer ${WAHOO_ACCESS_TOKEN}" \
"https://api.wahooligan.com/v1/workouts/WORKOUT_ID"
```
The FIT file URL lives at `workout_summary.file.url`.
### Download a FIT File
```bash
curl -L -o ride.fit "$FIT_URL"
```
The CDN doesn't require auth and doesn't count against your API rate limit.
### Get User Profile
```bash
curl -s -H "Authorization: Bearer ${WAHOO_ACCESS_TOKEN}" \
"https://api.wahooligan.com/v1/user"
```
Returns `{ id, height, weight, first, last, email, birth, gender, created_at, updated_at }`. Height and weight are returned as decimal strings.
### Refresh Access Token
```bash
bash {baseDir}/scripts/refresh_token.sh
```
The Python OAuth helper auto-refreshes on 401 if a `refresh_token` is on file. The shell helper is for manual/cron use.
```bash
curl -s -X POST https://api.wahooligan.com/oauth/token \
-d client_id="${WAHOO_CLIENT_ID}" \
-d client_secret="${WAHOO_CLIENT_SECRET}" \
-d grant_type=refresh_token \
-d refresh_token="${WAHOO_REFRESH_TOKEN}"
```
The new access token does not invalidate the old one until you successfully use it (Wahoo allows up to 10 unrevoked access tokens per user as of Jan 2026).
## Common Data Fields
`workout_summary` includes:
- `ascent_accum` — total elevation gain (m)
- `cadence_avg` — average cadence (rpm)
- `calories_accum` — kcal
- `distance_accum` — distance (m)
- `duration_active_accum` / `duration_paused_accum` / `duration_total_accum` — seconds
- `heart_rate_avg` — bpm
- `power_avg` — average power (W)
- `power_bike_np_last` — normalized power
- `power_bike_tss_last` — Training Stress Score
- `speed_avg` — m/s
- `work_accum` — total work (J)
- `time_zone` — IANA tz
- `file.url` — FIT file URL (CDN)
All decimal values are returned as strings — cast before math.
`workout_type_id` is an integer enum whose mapping isn't published. Empirically observed: cycling rides come back as `0`. Treat as opaque and key off the FIT `sport`/`sub_sport` fields (parsed into `workouts.fit_*` columns) when you need to filter by activity type.
## Rate Limits
| Tier | per 5 min | per hour | per day |
|------|-----------|----------|---------|
| Sandbox | 25 | 100 | 250 |
| Production | 200 | 1,000 | 5,000 |
Headers: `X-RateLimit-Limit`, `X-RateLimit-Remaining`, `X-RateLimit-Reset`. The sync pipeline backs off automatically on 429.
## Tips
- Convert m → mi: divide by 1609.34
- Convert m/s → mph: multiply by 2.237
- Decimal strings: `float(workout_summary["power_avg"])` before any math
- The list endpoint omits `workout_summary` — always hit detail (`/v1/workouts/:id`) to get FIT URL
## Error Handling
| Status | Meaning | Action |
|--------|---------|--------|
| 401 | Access token expired/invalid | Run `refresh_token.sh` or let `wahoo_auth.py` auto-refresh |
| 403 | Scope insufficient | Re-authorize with the missing scope |
| 429 | Rate limit hit | Wait until `X-RateLimit-Reset` |
| 404 | Workout not found / not yours | Confirm ID + ownership |
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes
Use when implementing any feature or bugfix, before writing implementation code