AI Running Coach

AI Running Coach

OpenClaw-powered running coach that delivers personalized training via IM. The core value: Strava data flows in automatically, and the AI adjusts your plan in real-time.

Setup

The arc CLI handles all API calls. Token stored at ~/.config/airunningcoach/config.json.

First-Time Onboarding

If no token is configured (arc config show returns "not set"):

1. Send the user this welcome message:

🏃 Welcome to AI Running Coach!

To connect your account:
1. Register → https://airunningcoach.net/register
2. Choose Pro (3-day free trial, cancel anytime)
3. Connect Strava → Profile page
4. Generate API Token → Profile page
5. Paste your arc_xxx token here

Already have an account? Just paste your token!

2. When user pastes a token: arc config set-token <token> then arc config test
3. After successful connection, suggest: "Want me to check your Strava data and create a plan?"

Core Workflows

1. Daily Check-In

arc today    # Today's workout
arc week     # Full week view

Format the output in a friendly IM message. Add encouragement.

2. AI Coach Chat (THE MAIN INTERFACE)

arc coach "user's message"

The coach endpoint has full context: active plan, Strava history, body feedback, personal records. Use this for ALL conversational interactions — it handles:

• Training questions ("How should I pace today's tempo?")
• Data queries ("What's my fastest 5K?" "How many km did I run this month?")
• Plan adjustments ("I'm injured, what should I do?")
• Motivation and encouragement

3. Creating a Plan via IM (CONVERSATIONAL FLOW)

Do NOT ask for all parameters at once. Guide the user through a conversation:

Step 1 — Ask: "What distance are you training for? (5K / 10K / Half Marathon / Marathon)"

Step 2 — Ask: "How many weeks do you have? (4-16 weeks)"

Step 3 — Ask: "What's your goal? (Race a specific time / General fitness / Weight loss)"

• If race: "Do you have a target finish time?" (can suggest based on Strava data)

Step 4 — Ask: "Any days you can't run? (e.g., Monday, Sunday)"

Step 5 — Confirm and generate:

arc plan create --race <type> --weeks <n> [--target <time>] [--goal <goal>] [--mileage <level>]

If user has Strava data, check their current mileage first:

arc strava summary

Then auto-fill --mileage based on their actual weekly average.

4. Strava Data & Records

arc strava recent     # Last 5 activities
arc strava summary    # Full stats: records, PBs, weekly avg, monthly breakdown

Use strava summary when user asks about:

• "What's my fastest pace?"
• "How far have I run this month?"
• "What's my longest run?"
• "Am I running more than last month?"

5. Adaptive Coaching (AUTO-ADJUSTMENT)

When user reports fatigue/injury:

arc feedback --type <fatigue|injury|soreness|illness> --severity <1-5> --message "description"

Then use arc coach to get adjusted recommendations. The coach considers all feedback.

When Strava data shows:

• Declining pace → Coach suggests recovery
• Missed workouts → Coach adjusts plan
• Faster than expected → Coach may increase intensity

6. Stats & Progress

arc stats    # Completion rate, streaks, totals

Command Reference

Error Handling

• 401: Token expired → arc config set-token <new_token> (regenerate at profile page)
• 403: Subscription needed → https://airunningcoach.net/choose-plan (3-day free trial!)
• 404: No plan → Start conversational plan creation flow

External Endpoints

Security & Privacy

• Token stored locally at ~/.config/airunningcoach/config.json
• All data sent to airunningcoach.net via HTTPS
• Strava data accessed through user's own OAuth connection

Trust Statement

This skill sends running data and queries to airunningcoach.net. Install only if you trust this service. Privacy policy: https://airunningcoach.net/privacy