autonomous-tests

Dynamic Context

• Args: $ARGUMENTS
• Branch: !git branch --show-current
• Unstaged: !git diff --stat HEAD 2>/dev/null | tail -5
• Staged: !git diff --cached --stat 2>/dev/null | tail -5
• Commits: !git log --oneline -5 2>/dev/null
• Docker: !docker compose ps 2>/dev/null | head -10 || echo "No docker-compose found"
• Config: !test -f .claude/autonomous-tests.json && echo "YES" || echo "NO -- first run"
• Capabilities: !python3 -c "import json;c=json.load(open('.claude/autonomous-tests.json'));caps=c.get('capabilities',{});mcps=len(caps.get('dockerMcps',[]));ab='Y' if caps.get('frontendTesting',{}).get('agentBrowser') else 'N';pw='Y' if caps.get('frontendTesting',{}).get('playwright') else 'N';cd='Y' if caps.get('frontendTesting',{}).get('chromeDevtools') else 'N';ec=sum(1 for s in c.get('externalServices',[]) if s.get('cli',{}).get('available'));print(f'MCPs:{mcps} agent-browser:{ab} playwright:{pw} chrome-devtools:{cd} ext-clis:{ec} scanned:{caps.get(\"lastScanned\",\"never\")}')" 2>/dev/null || echo "NOT SCANNED"

Role

Project-agnostic autonomous test runner. Analyzes code changes, auto-detects project capabilities, generates comprehensive test plans covering integration tests (curl-based API testing) and E2E tests (browser-based user flows), executes test suites via subagents, and documents findings for the test-fix-retest cycle.

Test Taxonomy

This skill generates three types of tests. Read references/test-taxonomy.md for full definitions:

• Integration Tests: API-level via curl. Security-focused request/response analysis. Classification: integration/api.