qalisted

# QA Pass Systematically test a web application and report findings. Inspired by [gstack /ai:qa](https://github.com/garrytan/gstack) — stripped of gstack-specific tooling. Browser/console output is large; quote only salient lines back into this context. Filter console logs with patterns instead of dumping all output. See [`context-discipline.mini.md`](../../../standards/rules/context-discipline.mini.md). ## Run mode - **Claude Code (preferred):** delegate the full QA pass to the `qa-runner` subagent via the Task tool with `subagent_type=qa-runner`. Pass: target URL, environment (`local`/`staging`), critical flows, tier (`quick`/`standard`/`exhaustive`). The subagent drives Playwright / CLI checks so the large browser and console output stays out of this context; it returns the markdown report below. Surface that report to the user, then run the fix loop here if asked. - **Browser-MCP interactive QA, or hosts without subagents:** run the process below inline. The process below is the canonical source of truth — `qa-runner`'s system prompt mirrors it. ## Process 1. **Clarify scope** — URL, environment (local/staging), and critical user flows. 2. **Smoke test** — load the app, check console errors, verify primary flows work. 3. **Deep test** — edge cases, form validation, auth boundaries, error states. 4. **Report** each issue with: severity (critical/high/medium/low), steps to reproduce, expected vs actual. 5. **Fix loop** (only if asked) — fix one issue, commit, re-test