ai-regression-testinglisted

# AI Regression Testing Add tests that catch the failures AI agents commonly miss when the same model writes code and reviews its own assumptions. ## Contract Inputs: - Bug report, code diff, API change, route, component, feature flag, or sandbox path - Existing test commands and project conventions - Optional known production incident or regression id Outputs: - Regression test plan - New or updated tests when implementation is requested - Parity checklist for sandbox, mock, feature-flag, and production paths - Verification commands and results Creates/Modifies: - Test files, fixtures, mocks, and minimal helpers when asked to implement - Application code only when the user also asks to fix the failing behavior External Side Effects: - None by default - Do not hit production services; use sandbox, fixtures, local test databases, or mocked providers Confirmation Required: - Before changing production data, external accounts, CI settings, or destructive test fixtures Delegates To: - `testing-expert` for general testing strategy - `react-testing-library` for React component tests - `playwright-e2e-init` or `e2e-testing` for browser workflows - `debug` when the root cause is still unknown ## When to Use - AI modified API routes, backend logic, serializers, or response schemas. - A bug was fixed and must not reappear. - A feature has sandbox/mock and production paths. - A model-generated review says the change is correct but no test proves it. - Build and lint pass