| name | browser-qa |
| description | Use when acting as `browser_debugger` to gather browser evidence, run the governed QA passes, and report PASS/FAIL/BLOCKED outcomes for UI-facing work. Do not use for application-code edits. |
Browser QA
Use this skill when acting as the browser_debugger role for a multi-agent ticket.
Read First
/hyperopen/AGENTS.md/hyperopen/docs/MULTI_AGENT.md/hyperopen/docs/agent-guides/browser-qa.md/hyperopen/docs/FRONTEND.md/hyperopen/docs/runbooks/browser-live-inspection.md
Job
- Reproduce the ticket in the browser when browser QA is required by the spec.
- Use the existing browser-inspection MCP server and artifacts.
- Report PASS, FAIL, or BLOCKED for all required passes, or emit an explicit skip when browser QA is not required.
- End browser work by stopping active sessions. Use
npm run browser:cleanup unless a narrower single-session stop is more appropriate.
Guardrails
- Do not edit application code.
- Use
/hyperopen/tmp/browser-inspection/** for transient evidence.
- Write
/hyperopen/docs/qa/** only when the workflow calls for a durable UI note.
Expected Outputs
browser-report.json- Optional QA note under
/hyperopen/docs/qa/
