run-quality-gate

name	run-quality-gate
description	Run full quality gate: linting, type checks, unit tests, and integration tests — all in parallel.
disable-model-invocation	false

Run Quality Gate

Runs the full verification pipeline for high-confidence changes. All gates run in parallel via background agents for maximum speed.

Project assumptions — detect and adapt before running

The gate commands below assume a Python-backend (uv, src/, tests/) plus frontend/ (Jest + Playwright) layout. Before running, detect the actual project and skip or adapt any gate that does not apply:

• No uv.lock → drop the uv run prefix (use plain pytest/ruff/pyright, or the project's package manager).
• No frontend/ dir → skip Gates 3, 5, 6 (TypeScript, Jest, Playwright).
• Source dir isn't src/ → substitute the real path; detect the test runner via the same table /run-unit-tests uses.
• No Playwright / no test:e2e:safe script → skip Gate 6.

Run only the gates that match the project. A gate skipped because its tooling is absent is a PASS, not a failure.

Usage

/run-quality-gate

Execution Strategy

IMPORTANT: Run all 6 gates in parallel as background agents in a single message. Do not run them sequentially. Each gate is independent — spawn all 6 agents at once and wait for results.

Gate 1: Lint/format (Python)

# Auto-detect: if project has uv.lock, use uv run
uv run ruff check --fix src/ tests/ && uv run ruff format src/ tests/
# If ruff --fix made changes, stage them

Agent prompt: "Run ruff check --fix and ruff format on src/ and tests/. Stage any auto-fixed files. Report errors that couldn't be auto-fixed."

Gate 2: Python type checking

uv run pyright src/

Agent prompt: "Run pyright on src/. Report all type errors with file:line locations."

Gate 3: TypeScript type checking

cd frontend && npx tsc --noEmit

Agent prompt: "Run tsc --noEmit in the frontend/ directory. Report all type errors."

Gate 4: Python unit tests

uv run pytest tests/ -q

Agent prompt: "Run the full pytest suite. Report pass/fail counts and any failures with tracebacks."

Gate 5: Frontend unit tests (Jest)

cd frontend && npx jest --no-coverage

Agent prompt: "Run the full Jest test suite in frontend/. Report pass/fail counts and any failures."

Gate 6: E2E tests (Playwright)

cd frontend && npm run test:e2e:safe

This runs playwright test --grep-invert @mutation (read-only, skips mutation tests). Agent prompt: "Run Playwright E2E tests via npm run test:e2e:safe in frontend/. Report pass/fail counts and any failures."

After All Gates Complete

1. Collect results from all 6 agents
2. Present the summary table
3. If ANY gate failed, fix the errors — do not just report them
4. Re-run only the failed gates after fixing
5. Repeat until all gates pass

Enforcement

• Any failing gate means overall failure.
• Fix ALL errors AND warnings — lint errors, lint warnings, type errors, and test failures must all be fixed regardless of when they were introduced. "Pre-existing" is not a valid reason to skip, and a warning is not a valid thing to leave behind. A gate passes only at zero errors and zero warnings.
• Never silence to pass. No inline eslint-disable / @ts-ignore / @ts-expect-error, no .skip / .only, no skipping failing suites. Fix the underlying cause. If a lint rule is genuinely wrong for this codebase, change the lint config deliberately and say so — don't suppress inline.
• Do not report errors and move on — fix them before claiming the gate passed.

Output Format

## Quality Gate Results

| Gate | Result | Details |
|------|--------|---------|
| Ruff lint/format | PASS/FAIL | N errors |
| Pyright | PASS/FAIL | N errors |
| TypeScript | PASS/FAIL | N errors |
| Pytest | PASS/FAIL | N passed, N failed |
| Jest | PASS/FAIL | N passed, N failed |
| Playwright E2E | PASS/FAIL | N passed, N failed |
| **Final verdict** | **PASS/FAIL** | |