The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes local utility scripts, specifically validate-yaml.py and search-yaml.py, to ensure the integrity of generated documentation and to manage the project manifest. These operations are restricted to the local environment for validation and metadata management.
[PROMPT_INJECTION]: The skill processes source code (untrusted external data) to generate documentation, which presents a surface for indirect prompt injection.
Ingestion points: Reads files from source_root and source_files provided by the user.
Boundary markers: Includes strict instructions to focus on 'signatures, type definitions, and default values' and explicitly forbids 'guessing' or 'fabricating' content.
Capability inventory: Executes validate-yaml.py and search-yaml.py; performs file writes to the docs/api/ directory.
Sanitization: Employs YAML validation scripts to verify the structure and required fields of the generated output.

cs-libdoc