Execution Quality Gate

Activation

1. •Activate during final gate steps.
2. •Activate on explicit $codex-execution-quality-gate.
3. •Run before saying work is complete.
4. •Activate on $pre-commit or "check before commit".
5. •Activate on $smart-test or "which tests to run".
6. •Activate on $suggest or "suggest improvements".
7. •Activate on $impact or "what will this affect".
8. •Activate on $quality-record to record quality trend snapshot.
9. •Activate on $quality-report to generate trend report.
10. •Activate on $ux-audit.
11. •Activate on $a11y-check.
12. •Activate on $lighthouse <url>.
13. •Activate on $e2e check.
14. •Activate on $e2e generate <url>.
15. •Activate on $e2e run.

Phase X Verification Order

Execution

1. •Run run_gate.py with project root.
2. •Run security_scan.py with project root.
3. •Optionally run bundle_check.py with project root.
4. •Optionally run tech_debt_scan.py with project root.
5. •Optionally run suggest_improvements.py after complex tasks.
6. •Optionally run predict_impact.py before high-risk edits.
7. •Optionally run quality_trend.py --record on periodic cadence.
8. •Optionally run ux_audit.py before UI handoff.
9. •Optionally run accessibility_check.py for public-facing surfaces.
10. •Optionally run lighthouse_audit.py before deploy (requires running server URL).
11. •Optionally run playwright_runner.py for E2E setup/check/run flows.
12. •Merge results and decide pass/fail.

Decision Rules

• •Completion allowed only if no blocking failures remain.
• •Blocking failures must be fixed, then checks rerun.
• •Warning-only outcomes may proceed with explicit warning to user.
• •Tech debt scan is advisory in MVP and does not block completion.
• •Improvement suggestions are advisory and do not block completion.
• •Impact prediction is advisory and does not block completion.
• •Quality trend reporting is periodic/advisory and does not block completion.
• •UX static audit is advisory and does not block completion.
• •Accessibility static checker is advisory and does not block completion.
• •Lighthouse wrapper is advisory and must fail gracefully when tool/server is unavailable.
• •Playwright wrapper is advisory and must fail gracefully when tool/setup is unavailable.

Script Paths

• •Windows: python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\<script>.py" --project-root <path>
• •macOS/Linux: python "$HOME/.codex/skills/codex-execution-quality-gate/scripts/<script>.py" --project-root <path>

Tech Debt Command

• •python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\tech_debt_scan.py" --project-root <path>

Pre-Commit Intelligence Command

• •python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\pre_commit_check.py" --project-root <path>

Smart Test Selector Command

• •python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\smart_test_selector.py" --project-root <path> --source staged

Improvement Suggester Command

• •python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\suggest_improvements.py" --project-root <path> --source last-commit
• •In proactive mode, run after complex tasks and present top 3 suggestions.

Impact Predictor Command

• •python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\predict_impact.py" --project-root <path> --files <file1,file2> --depth 2

Quality Trend Commands

• •Record: python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\quality_trend.py" --project-root <path> --record
• •Report: python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\quality_trend.py" --project-root <path> --report --days 30

UX Audit Command

• •python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\ux_audit.py" --project-root <path>

Accessibility Check Command

• •python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\accessibility_check.py" --project-root <path> --level AA

Lighthouse Audit Command

• •python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\lighthouse_audit.py" --url <http://localhost:3000> --device mobile --runs 1
• •Requires a running URL and Lighthouse availability through npx.

Playwright Runner Commands

• •Check setup: python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\playwright_runner.py" --project-root <path> --mode check
• •Generate skeleton: python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\playwright_runner.py" --project-root <path> --mode generate --url <http://localhost:3000/page>
• •Run tests: python "$env:USERPROFILE\.codex\skills\codex-execution-quality-gate\scripts\playwright_runner.py" --project-root <path> --mode run --browser chromium

Reference Files

• •references/gate-policy.md: blocking vs warning rules for gate decisions.
• •references/improvement-suggester-spec.md: post-task suggestion behavior and presentation protocol.
• •references/impact-predictor-spec.md: pre-edit blast-radius analysis guidance.
• •references/quality-trend-spec.md: periodic quality trend workflow and interpretation.
• •references/ux-audit-spec.md: static UX audit checks and scoring.
• •references/accessibility-check-spec.md: WCAG static checks and compliance scoring.
• •references/lighthouse-audit-spec.md: Lighthouse wrapper behavior and graceful fallback.
• •references/playwright-runner-spec.md: Playwright check/generate/run behavior.

Override

If user says skip gate or force complete, comply and warn: "Quality gate skipped. Lint/test/security status is unknown."

Output Handling

For each script:

1. •capture output
2. •summarize errors/warnings/passes
3. •ask whether to fix blocking errors when present
4. •rerun after fixes to verify