The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several shell commands, including git diff, grep, find, sed, and awk, to audit the repository for documentation drift and verify path references. These commands are essential to the skill's primary function of maintaining documentation accuracy and detecting staleness.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests and processes untrusted data from the project repository to drive documentation updates.
Ingestion points: The skill reads multiple repository files, including CLAUDE.md, .claude/rules/*.md, README.md, docs/**/*.md, and package.json (as described in rules/drift-detection.md).
Boundary markers: No explicit boundary markers or "ignore embedded instructions" warnings are utilized when reading these files.
Capability inventory: The skill possesses capabilities for file modification (Edit, Write) and shell command execution (git, grep, find, sed, awk, npx).
Sanitization: No sanitization or validation of the ingested markdown content is performed before processing it to generate new documentation or execute audit commands.
[EXTERNAL_DOWNLOADS]: The skill recommends and uses third-party tools via npx for documentation linting and link checking, specifically markdownlint-cli2, alex, and lychee. These are well-known services and tools commonly used in development workflows for quality assurance.

docs