alpacalyzer-algo-trader: Skill for Claude Code

Validate End-to-End Skill

When to Use

After making changes to the trading pipeline (scanners, agents, strategies, execution), validate that the full flow still works by running a dry-run cycle.

How to Validate

uv run alpacalyzer --analyze --dry-run --json 2>/dev/null

This runs one analysis cycle and outputs structured JSON to stdout:

{
  "status": "ok",
  "mode": "dry_run",
  "tickers_scanned": ["AAPL", "TSLA", "NVDA"],
  "opportunities": 3,
  "strategies_generated": 2,
  "signals_queued": 2,
  "errors": []
}

What to Check

1. status should be "ok" — any other value means something broke
2. errors should be empty — non-empty means a component failed
3. opportunities > 0 — if zero, scanners may be broken
4. strategies_generated > 0 — if zero, agents may be broken

With Specific Tickers

uv run alpacalyzer --analyze --dry-run --json --tickers AAPL,MSFT 2>/dev/null

Interpreting Failures

• "status": "error" with "errors": ["scanner timeout"] → Scanner issue
• opportunities: 0 → Check scanner adapters and API keys
• strategies_generated: 0 with opportunities > 0 → Agent/LLM issue
• Any Python traceback on stderr → Code error (the 2>/dev/null hides stderr; remove it to debug)

Integration with CI

This is not wired into CI (it requires API keys). Use it locally after making pipeline changes.