Agent Groundcheck: CI tool to catch broken coding-agent instructions
Agent Groundcheck scans a repository’s base and head revisions to find instruction claims that become invalid after a change. It classifies findings into new, unproven, existing, and fixed, only blocking CI on newly introduced failures. This helps developers maintain accurate documentation for AI coding agents without cleaning historic debt. It’s a lightweight CLI that integrates easily into CI pipelines.
View on GitHub →yo4e/agent-groundcheck