test-qa — Story-driven CRUD, not a monkey test

Degree of freedom: MIXED — story generation and severity [HIGH freedom]; session, anti-stall, and CRUD verify-after-refresh [LOW freedom — run exactly]. Driver is playwright-cli, never Playwright MCP. Guest-vs-authed wander is test-exploratory — do not reclaim it.

Full QA of a live webapp as a senior QA engineer preparing a production release: user-story-driven coverage, CRUD, data-pipeline integrity, UX, and edge cases. This is not a page-navigation monkey test. Your turn ends after Phase 5 with every discovered route crawled and every CRUD entity proven after refresh — not after the crawl with CRUD still owed, and not with an auth-required page silently skipped instead of marked BLOCKED.

Read protocol-browser-anti-stall before any browser step (Rule 0: manual & headed). Also references/playwright-session-coordination.md in that folder.

This skill vs neighbors