Agent Browser QA Reporting

Use this skill to turn agent-browser dogfood, pre-QA validation, focused reproduction, or fix-verification output into a clear QA report and tester handoff.

Process

1. Confirm the reporting contract: tested scope, target environment, deployment/build, browser or device set, roles, acceptance criteria, artifacts, and who will consume the report.
2. Before any command-sensitive agent-browser work, check the installed command surface first:
   • agent-browser --version
   • agent-browser --help
3. If the installed CLI version supports versioned or local guidance such as agent-browser skills get core or agent-browser skills get dogfood, load it and treat it as the source of truth for exact commands, flags, artifact paths, and behavior. Older versions may not provide those commands.
4. Treat static CLI examples as patterns only.
5. Build the coverage matrix before writing conclusions. Mark each area as covered, partially covered, blocked, untested, or not applicable, and state why.
6. Normalize findings with type, severity, priority, confidence, reproducibility, evidence, and next verification. Do not file a product bug from tooling failure alone.
7. Redact secrets, PII, tokens, cookies, auth state, authorization headers, tenant-sensitive data, and sensitive payloads before sharing the report or evidence.
8. Separate facts from hypotheses. Keep suspected owners and root-cause notes provisional unless implementation evidence confirms them.

Output Standard