verification-loop

Automating code verification before a pull request requires running build, type, lint, test, and security checks. This skill coordinates those six phases end-to-end and enforces a ready-to-PR gate to ensure only compliant code reaches review.

Running lint and security scans during feature completion is best handled by an automated verification loop. This skill applies those checks alongside build and type validation, outputting a detailed report of any failures before you finalize changes.

Yes, the verification loop works without external CI-CD dependencies. It operates autonomously within Claude Code sessions to enforce quality assurance gates, requiring no additional components or external libraries to execute its six verification phases.

You can use this automated testing workflow for refactoring existing code to ensure no regressions occur. It validates types, builds, and tests across the modified diff, generating a verification report that confirms the refactored code meets quality gates.

A diff check is needed in the verification process to isolate and validate only the modified code changes. This phase ensures the specific build, type, lint, and security impacts of your recent edits are caught before merging.

A limitation of enforcing quality gates locally in Claude Code is that it validates the session's current state rather than the remote CI pipeline. You must still run your external CI-CD platform to catch environment-specific deployment or integration failures.