Enforces a rigorous evidence-based verification process to ensure code changes and tasks are truly successful before they are marked as complete.
This skill transforms Claude's behavior by instilling a verify-first mentality, preventing premature claims of success during the development lifecycle. It mandates that the AI agent executes actual commands—such as test runners or build scripts—and explicitly quotes the output before reporting a task as finished. By eliminating assumptions and 'should work' reasoning, it significantly reduces regression rates and improves the reliability of AI-assisted engineering workflows, especially when integrated with the Ito task management system.
Key Features
011 GitHub stars
02Pre-commit and pre-archive validation protocols
03Evidence-based verification via direct output quoting
04Detection and prevention of 'low risk' assumption traps
05Integration with Ito workflow task completion
06Mandatory command execution before success reporting
Use Cases
01Verifying build success after adding new dependencies to a project
02Ensuring unit tests pass after a bug fix before closing a ticket
03Confirming acceptance criteria are met for complex engineering tasks