agent-tooling

Verify a change in a real HEADLESS browser: figure out what to look for, drive headless Chromium with Playwright, and either confirm a bug still reproduces or verify it's fixed AND that nothing on the risky adjacent paths broke. Use whenever the user wants to "check this fix in the browser", "confirm/repro this bug headlessly", "test the UI locally", "did the fix work and break anything", "run a headless browser test", "verify the arguments tab / voting / a page works", or hands you specific things to test in the running app. Prefer this over a screenshot or a code-only review when the question is "does it actually behave correctly in a browser?". Lighter than a full PR gate; pairs with pr-check's local-verification step.

Full PR quality-gate: scope the diff, run the project's tests, convene a type-matched panel of expert review agents with a cross-review round, conditional security review and targeted local verification, then a final go / merge-with-fixes / needs-changes verdict plus a call on whether a human staging test is warranted. Use whenever the user asks to "check this PR", "quality-check before merge", "run the PR gate", "is this ready to merge", "vet this branch", "/pr-check", or wants a consolidated go/no-go on a branch or PR — even if they don't name a specific check. Prefer this over running a code review or the test suite alone when the ask is "is this good to merge?".

5 hooks across 2 events