Summary

• Add a bridge layer (server/bridge/checker.ts) that spawns an external Python health checker process and caches the JSON results in data/last-check.json
• Add 4 health API endpoints (/api/health/results, /api/health/status, /api/health/run, /api/health/skill/:name) for the frontend to consume
• Add a new "健康" (Health) tab in the Web UI with a full dashboard showing per-skill health scores across 4 dimensions: code quality, git sync, runtime health, and activity
• The Python checker is optional — if not present, the UI degrades gracefully (shows "no data, run a check" prompt)

New files

Modified files

Architecture

Browser → Fastify Server (TS)
           ├── /api/health/results  → read data/last-check.json (cached)
           ├── /api/health/run      → spawn python3 checker/main.py --json
           └── /api/health/status   → check if Python process is running

Python checker (optional, user-provided):
  checker/main.py --json → data/last-check.json
  - 4-dimension scoring: code quality (flake8/black/bandit) + git sync + runtime + activity
  - Each dimension 0-25 points, total 0-100

Screenshots

The health dashboard displays:

• Summary cards (total checked / healthy / warning / critical / timing)
• Per-skill table with score badges, category badges, score bars, and staleness indicators
• One-click "Run Check" button with polling progress

Test plan

• npm run dev → verify all existing views (Skills, 相似检测, 仪表盘) still work
• Click "健康" tab → shows "no data" prompt (no Python checker yet)
• Place a Python checker at checker/main.py with --json support → click "Run Check" → verify results display
• Verify /api/ping still returns { status: 'ok' }
• Verify npm run build completes without TypeScript errors