agents-md-improver

To audit AGENTS.md files, this Skill analyzes the effective chain of AGENTS.md and AGENTS.override.md files to identify shadowed or missing instructions, ensuring your AI agents follow the correct project-specific guidance.

Your AI agent might ignore project-specific rules due to gaps or inconsistencies in the repository-scoped Codex instruction chains. This Skill audits AGENTS.md files to resolve these issues and generate navigation maps for complex repositories.

The best way to generate a navigation map is by using this Skill to maintain code_map.md files, which directly improves AI navigation in large, multi-module codebases by optimizing repository guidance.

Cleaning up stale commands in AGENTS.override.md requires auditing the repository-scoped Codex instruction chains. This Skill validates and applies instruction updates to resolve inconsistencies and remove outdated commands.

Yes, optimizing Codex guidance requires both read and write access to repository configuration files. This access is necessary to validate the instruction chains and apply the proposed instruction updates.

You should avoid an AGENTS.md instruction audit if you lack write access to repository configuration files, as this Skill needs those permissions to validate and apply the necessary instruction updates.