The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use shell utilities including ls, find, grep, and test -e to audit repository structure, discover components, and verify the validity of file paths referenced in the documentation.
[COMMAND_EXECUTION]: The workflow includes directives to manage documentation symlinks using rm and ln -s, specifically to ensure that CLAUDE.md points to AGENTS.md and contains no unique, unmerged content.
[PROMPT_INJECTION]: The skill processes untrusted input from AGENTS.md files within the repository, creating a surface for indirect prompt injection. Ingestion points: Content from AGENTS.md is read and analyzed during the audit. Boundary markers: The instructions include explicit safeguards such as 'Ask the user. Never autoresolve' and 'Produce the diff plan. Do not apply edits.' Capability inventory: File system inspection (ls, find, test, grep) and documentation management (ln, rm). Sanitization: No formal sanitization of the input documentation is specified beyond the directive for human review of the plan.
[SAFE]: The skill's behavior is entirely consistent with its stated purpose of documentation maintenance. It does not perform network operations, access sensitive credentials, or employ obfuscation techniques.

agents-md