| Primary focus | Product-specific errors in live PR and release builds | Implementation issues in the pull request | Screenshot, component, and DOM changes | Runtime errors, latency, and incidents |
|---|
| What it inspects | Billing, auth, permissions, AI behavior, and workflows | Code diffs, tests, static analysis, and repository context | Pixels, layouts, components, and visual baselines | Logs, traces, metrics, and production events |
|---|
| Issues it finds | Wrong prices, exposed routes, broken roles, changed AI answers, and workflow regressions | Bugs, security problems, style issues, and test gaps visible in the code | Layout breaks, styling drift, and unexpected UI changes | Exceptions, slow requests, outages, and service degradation |
|---|
| Review output | Notary Flags, evidence, Product State Hash, release notes, and a GitHub check | Inline comments, summaries, and suggested code changes | Visual diffs, snapshots, and baseline approvals | Alerts, traces, dashboards, and incident timelines |
|---|
| When it helps | Before merge or release, against the deployed PR or release build | During pull-request review, from repository context | During UI review and continuous integration | After deployment, while the system is running |
|---|