Goal
Deliver a complete Playwright TypeScript E2E suite covering every UI phase of the new React frontend, runnable in CI at $0 cost via the Epic 8 fake-LLM provider.
Delivery is delegated to the user's existing PW agent team at ~/.claude/test-agents/:
- Entry point:
pw-team (single orchestrator — Analyze → Scaffold → Write → Review)
- Workers:
pw-framework-builder, pw-dom-analyzer, pw-test-writer, pw-reviewer
- Conventions: established 4-layer architecture (Specs → Steps → Page objects → Utilities)
- Shared state:
.pw-agents/project-state.md (PW team owns this file)
RTIA does not prescribe the framework structure, test layout, or fixture pattern — those are the PW team's domain.
User Stories
Depends On
Out of Scope
- Test architecture, page-object layout, fixture format — owned by the PW agent team.
- Smoke testing against a real LLM (Ollama / Gemini) — covered ad-hoc by maintainer, not by this Epic.
- Visual regression (screenshot diffing) — can be added later if needed.
Definition of Done
Goal
Deliver a complete Playwright TypeScript E2E suite covering every UI phase of the new React frontend, runnable in CI at $0 cost via the Epic 8 fake-LLM provider.
Delivery is delegated to the user's existing PW agent team at
~/.claude/test-agents/:pw-team(single orchestrator — Analyze → Scaffold → Write → Review)pw-framework-builder,pw-dom-analyzer,pw-test-writer,pw-reviewer.pw-agents/project-state.md(PW team owns this file)RTIA does not prescribe the framework structure, test layout, or fixture pattern — those are the PW team's domain.
User Stories
Depends On
Out of Scope
Definition of Done
ThreadStatusvalue (IDLE,PAUSED_PO,PAUSED_REVIEW,DONE,DONE_SPLIT,ERROR) has at least one spec exercising it.RTIA_LLM_PROVIDER=fake(zero API spend in CI).pw-reviewerreview pass reports no Critical or Major findings..pw-agents/project-state.mdcommitted to the repo..github/workflows/e2e.ymlruns the suite on every PR and uploadsplaywright-report/as an artifact.