The Agent Skills Directory

[SAFE]: No security issues detected. The skill outlines a documentation update workflow using standard local tools and incorporates human-in-the-loop verification before applying any file modifications.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing external content from Git diffs and MDX files. This is mitigated by mandatory user confirmation.
Ingestion points: git diff output and .mdx files
Boundary markers: None
Capability inventory: Execution of git and pnpm commands and file system write access
Sanitization: User review and confirmation of all changes before execution

update-docs