The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses dynamic context placeholders in the SKILL.md file to fetch repository status (branch name, commit count) at load time. These commands are limited to benign, read-only git operations.
[PROMPT_INJECTION]: The skill processes untrusted commit messages to detect missing documentation, creating a surface for indirect prompt injection. The risk is considered negligible as the impact is confined to local documentation skeleton generation. * Ingestion points: Commit history is read via git log and the git-derive-docs.sh script. * Boundary markers: None identified. * Capability inventory: The agent has access to Write, Edit, and Bash tools to modify repository documentation. * Sanitization: No explicit sanitization of commit messages is performed before processing.

git-derive-docs