The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use standard system commands such as ls and find to inspect the repository's directory structure for context gathering.
[PROMPT_INJECTION]: The skill processes content from untrusted repository files (e.g., README.md, package.json) to generate the guidelines. This represents a surface for indirect prompt injection where malicious content in those files could influence the generated documentation, though the risk is limited to the documentation's content.
Ingestion points: Reads project files including README*, CONTRIBUTING*, DEVELOPMENT*, Makefile, pyproject.toml, and package.json.
Boundary markers: Not explicitly defined in instructions.
Capability inventory: Limited to writing the AGENTS.md file to the local filesystem.
Sanitization: None mentioned; relies on the agent's internal summarization logic.

init