DawnCoach Phase 1 Implementation - Issues #7, #8, #9, #10

session-transcript.md

DawnCoach Phase 1 Implementation Session

Session Summary

This session completed Phase 1 (Foundation) of DawnCoach, implementing the AI chat interface with Strava training context awareness.

Issues Completed

Issue #7: Fetch expanded Strava activity history (30-90 days)

• Added /api/training-context endpoint
• Fetches activities from Strava API with pagination (up to 1000 activities)
• Aggregates weekly metrics: distance, time, elevation, activity count
• Calculates training trends (increasing/stable/decreasing)
• Identifies primary activity type
• 5-minute cache to reduce API calls

Issue #8: Build training context prompt for Claude

• Created buildSystemPrompt() function for dynamic prompts
• Includes athlete training summary in system prompt
• Shows weekly breakdown (last 4 weeks)
• Handles missing data gracefully

Issue #9: Create chat UI component

• Chat container with message list
• User/assistant message styling (Strava orange theme)
• Loading indicator with animated dots
• Auto-scroll to latest message
• Error handling with user feedback

Issue #10: Add navigation between Activity View and Coach views

• Tab navigation UI
• Activity View shows latest Strava activity
• Coach View shows full-height chat interface
• URL hash deep linking (#activity, #coach)
• State persists during tab switches

Files Modified

• main.go - Added training context types, endpoint, and dynamic system prompts
• public/index.html - Added chat UI, navigation tabs, and related JavaScript
• CLAUDE.md - Updated gh CLI usage to require --repo flag

Commits

1. 947bcad - Add expanded Strava activity history endpoint (Closes #7)
2. de06a2b - Add training context-aware system prompt for Claude (Closes #8)
3. 9979cdb - Add chat UI component for AI coach interaction (Closes #9)
4. 57741da - Add tab navigation between Activity and Coach views (Closes #10)

PR Created

PR #35: https://github.com/cconcannon/dawncoach/pull/35

Technical Details

Training Context Structure

type TrainingContext struct {
    TotalActivities  int
    WeeklyVolume     []WeekSummary
    RecentTrend      string // "increasing", "stable", "decreasing"
    PrimaryActivity  string
    LastActivityDate string
    TotalDistanceKm  float64
    TotalTimeHours   float64
    AvgWeeklyDistKm  float64
    AvgWeeklyTimeHrs float64
    DaysAnalyzed     int
}

API Endpoints Added

• GET /api/training-context?days=60 - Returns aggregated training data
• POST /api/chat - Now accepts optional training_context in request body

Frontend Features

• Alpine.js reactive data for chat state
• Conversation history support for multi-turn chats
• Deep linking via URL hash for shareable view links