agf-running-sit-tests

# Running System Integration Tests (SIT) Use this skill when: - A dev finished feature code + Unit tests and is about to enter code-review - A dev needs to verify a fix doesn't regress integration (API ↔ DB ↔ external) ## What SIT is — and is not **SIT** verifies that **independently-developed components compose correctly** — frontend ↔ backend ↔ DB ↔ external services. It is NOT: - Unit tests (already covered by `pytest` / `vitest` on the branch before SIT) - E2E tests (real browser + real user journey — run downstream after code-review) - UAT (business owner final sign-off — product-lead drives, downstream of E2E) You just wrote the Unit tests, so you have the clearest picture of the unit-vs-integration boundary. If a failure reproduces by running just the backend unit tests with mocks, it's a unit-level miss, not a SIT finding; fold it back into the unit suite rather than writing it up as a SIT defect. ## Pre-conditions - [ ] Feature branch is rebased onto `main` - [ ] All Unit tests passing on the branch (pytest / vitest green) + lint + typecheck green - [ ] AC 来源可访问：`docs/changes/<change>/tasks.md`（AC↔scenario 映射，[ADR-012](../../../docs/adr/012-spec-driven-change-folders.md)）；旧 feature fallback `docs/prd/[feature]-[date].md` - [ ] `.env.local` with SIT-mode flags configured (or `.env.sit` if a dedicated SIT config exists) - [ ] **前端 SIT 专属**：MSW mock 来自 orval 生成产物（`*.msw.ts`），非手写——mock 与 OpenAPI 契约同源（见 ADR-006 / `coding.md` 契约纪律） If any precondition fails: SendM...