The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted source code and extracts descriptions from docstrings and comments directly into the final OpenAPI specification. Ingestion points: Source files are read in scripts/generate_api_doc.py using Path.read_text() within the extract methods of the PythonExtractor, JSExtractor, and GoExtractor classes. Boundary markers: None. Extracted strings are placed directly into the summary and description fields of the OpenAPI output without specific delimiters or warnings to ignore embedded instructions. Capability inventory: The script performs file system reads to scan code and writes to save the specification, but it contains no network operations or shell command execution capabilities. Sanitization: The script does not sanitize or escape extracted docstrings or comments, which may allow malicious instructions embedded in the code to persist in the generated documentation.

api-doc-gen