verification-protocol

You verify software builds by applying a five-step gate function: IDENTIFY targets, RUN commands, READ outputs, VERIFY against expectations, and CLAIM completion. This standardized sequence captures reproducible evidence, preventing false positives in your engineering workflows.

Evidence-based verification is a process that standardizes the validation of task completion and claims. By requiring captured outputs, checklists, and documented blockers, it ensures correctness claims are backed by reproducible evidence, creating audit trails and increasing confidence in delivered work.

You create verification checklists by itemizing the steps required per task. The checklist captures the specific commands to run, expected outputs, and defined pass/fail criteria, ensuring standardized validation for unit and integration test suites.

Yes, you can perform partial verification for manual checks and flaky tests. The protocol provides specific edge-case guidance, requiring you to document blockers and record the manual verification steps taken, ensuring transparency even when full automation is not possible.

You validate TypeScript builds and deployment smoke tests by running the mandatory five-step gate. This requires you to identify the verification target, execute the build or smoke test, read the captured outputs, and verify the results against your pass criteria before claiming completion.

The best way to prove feature completion is using a gate function that demands reproducible evidence. By running itemized verification checklists and capturing actual outputs, you ensure the feature is validated against strict pass/fail criteria rather than subjective claims.