Vet – Local code review tool verifying AI coding agents

Summary: Vet is an open-source, fast, local code review tool that verifies AI coding agents by analyzing conversation history to ensure their outputs match requests. It detects silent failures like incomplete features, unrun tests, logic errors, and deviations from goals in full pull requests.

What it does

Vet reviews code generated by AI agents using prior interactions to identify issues such as half-implemented features, unexecuted tests, logic errors, and unhandled edge cases. It operates locally via CLI, CI, or as an agent skill without telemetry.

Who it's for

Developers using AI coding agents who need reliable verification of the agent’s code outputs and alignment with specified tasks.

Why it matters

It prevents silent failures by ensuring AI-generated code is accurate, complete, and consistent with user instructions, reducing unnoticed errors.