MCP Test Runner

Gives Claude local test execution and coverage feedback for Python and JavaScript projects. Exposes six tools: run_pytest, run_jest, run_single_test, parse_test_output, get_coverage_summary, and a health check. Built for the wayfinder codebase onboarding agent as the verification layer alongside mcp-repo-mapper and mcp-ast-explorer. Runs pytest with JSON output via pytest-json-report, Jest via npx with JSON flags, and normalizes both into a single TestRunResult schema. Includes subprocess timeouts and POSIX resource limits for CPU and memory. Coverage summaries work through pytest-cov. Reach for this when your agent needs to verify code claims against real test execution instead of guessing from static analysis.