Tool: Playwright E2E

개요: 실제 브라우저에서 사용자 흐름이 끝까지 동작하는지를 반복 가능한 스크립트로 증명합니다.

Proves

• A user-visible flow (open page → act → see result) works in a real browser engine
• The same flow keeps working later (repeatable, can run in CI)

Does not prove: visual polish, performance, backend internals.

Use when

• Frontend or BFF change touches a flow a user actually performs
• The orchestration lists E2E for this situation
• A manual smoke found a bug and you want it to stay fixed

Do not use when

• The project has no Playwright setup and the orchestration did not ask to add one (→ BLOCKED or browser-manual)
• The change is purely backend with no UI surface (→ backend-unit-test)

Inputs

Need	Source
Running app	dev server URL from the mission or project README
Existing specs	target repo's e2e/ / tests/ — follow their naming, fixtures, selectors
Selectors	prefer roles / test ids already used in the repo; do not invent a new selector strategy

Evidence format

Under ## Artifacts:

command: npx playwright test e2e/item-create.spec.ts
result:  3 passed (chromium)
files:   e2e/item-create.spec.ts (added)
trace:   playwright-report/ (if generated)

If a screenshot is the only readable proof, attach the path and say what it shows.

BLOCKED conditions

• App does not start / URL unreachable
• Login, captcha, or external auth blocks the flow and no test account is in context
• No Playwright in the project and no permission to add it

Current vs extension

Item	Status
Playwright with the repo's existing config	current where present
Adding Playwright to a project that lacks it	extension — requires the orchestration to say so
Visual regression snapshots	extension

Product mapping (how the agent runs this) lives in ../entrypoints/.