What problem does it solve? Running a project's full local quality gate by hand means remembering the right lint, typecheck, and test commands and wading through long raw output. This Skill reads the ordered command list from .harness/project.json, executes each command through a summarizing wrapper, and returns a compact pass/fail result with pytest totals and failed test node IDs. ## Core Features & Use Cases - Config-driven gate execution: Reads qa_gate_commands from .harness/project.json and runs them in order, stopping at the first failure. - Bounded failure summaries: The test_summary.py wrapper sanitizes output, extracts pytest counts and failed node IDs (capped list), and writes full logs to .harness/test-logs only on failure. - Pass marker recording: On success it records a QA-gate pass marker via the harness hook script so downstream PR workflows can verify the gate ran. - Use Case: Before opening a pull request, ask the agent to run the full quality gate; it executes make check and make test (or whatever is configured) and reports a one-line PASS or a focused failure summary. ## Quick Start Run the full project quality gate and tell me whether it passes before I open my pull request.
