Verified Completion Guard

About

The Verified Completion Guard implements a strict 'evidence-first' protocol for Claude Code, preventing premature or dishonest completion claims. It mandates the execution of fresh verification commands—such as tests, builds, and linters—and requires the agent to confirm success through actual command output rather than assumptions. By establishing a non-negotiable gate function before any PR, commit, or status update, it ensures high reliability, maintains developer trust, and prevents the shipment of broken or incomplete features.

Key Features

• Real-time exit code and failure count analysis
• Comprehensive 'Red-Green' verification for regression testing
• Rationalization prevention against 'should' or 'probably' language
• Mandatory command execution before any success claims
• 0 GitHub stars
• Independent verification of delegated agent reports and VCS diffs

Use Cases

• Verifying complex feature requirements against a line-by-line checklist
• Ensuring clean builds and linter passes before creating pull requests
• Validating bug fixes through reproduction and resolution testing cycles