The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run within scripts/document-claims.py to execute local git commands (e.g., ls-files, grep, log, rev-list) to verify the status and history of project files. The implementation follows security best practices by passing arguments as lists rather than using a shell environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from project documentation which could be manipulated to influence agent behavior.
Ingestion points: The skill reads README.md, AGENTS.md, and any project documents linked within them to verify their accuracy.
Boundary markers: No specific delimiters or warnings are used when processing document content in the provided Python script.
Capability inventory: The skill has the authority to write to documentation files, modify GitHub Actions workflow configurations (.github/workflows/checks.yml), and perform Git operations including committing, pushing, and opening pull requests.
Sanitization: The skill relies on natural language processing and basic string matching (regex) without explicit sanitization of embedded instructions in the ingested documents.

sync