verification-before-completion

Run your build verification command (e.g., npm run build, cargo build) and confirm the output shows success without errors. Attach the exact command output as evidence before stating completion; this blocks false claims when builds fail silently or partially.

Verification reads and confirms test output matches your claim—not just running the command. Execute the test command (npm test, pytest, npx playwright test), examine the results for failures or warnings, and document findings before declaring tests passed.

No. Verification-before-completion requires fresh evidence at each stage: build checks (npm run build), type checks (npx tsc --noEmit), unit tests, and E2E tests. Skipping gates allows broken code into release pipelines, defeating the verification model.

Treat warnings as evidence requiring interpretation. Read linter and type-checker output fully; determine if warnings block your claim or are acceptable for your release criteria. Document the decision with the actual output attached, so the gate is explicit and auditable.

The verification gate stops you. Completion claims require fresh command output and explicit result confirmation. Claims made without evidence are blocked, ensuring release readiness checks and PR sign-offs rest on verified facts, not assumptions.

Yes. Run and verify each build stage and test suite separately; attach outputs for all stages before claiming overall completion. The 5-step gate (identify, run, read, verify, claim) scales to complex pipelines—each component must produce fresh evidence.