The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands such as mkdir, mv, and cat to manage project metadata, archive completed workflow states, and load reference documentation. It also runs local Python scripts via tool hooks to enforce phase gates and verify edits.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface as it ingests and processes content from potentially untrusted user draft files, though the risk is mitigated by a rigid diagnostic-and-fix loop that focuses on specific identified issues.

writing-revise