Keeps your AGENTS.md files lean and up to date across monorepos and nested modules. It reads your package manifests, Makefiles, and CI configs to infer real commands instead of inventing them, then generates or repairs both root and module-level agent instructions while staying under 60 and 40 lines respectively. The progressive disclosure approach pushes detail into docs and skills rather than bloating the instructions. Handles the annoying bits like detecting module boundaries, preserving security warnings from existing files, and adapting skill references based on whether you have a local skill directory or rely on external ones. Useful when your agent context is missing, contradictory, or sprawling across a growing codebase.
npx skills add https://github.com/madteacher/mad-agents-skills --skill agents-md-generator