The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Python's subprocess module to coordinate internal data processing tasks across multiple utility scripts. Scripts such as batch_convert_templates.py, convert_to_json.py, and parse_markdown.py invoke subprocess.call() to run internal helpers like convert_formats.py and parse_formats.py. These operations are limited to local file system actions and use the current Python interpreter environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external project context provided by users, which represents a potential surface for indirect prompt injection. To mitigate this risk, the core instructions explicitly direct the agent to distinguish between confirmed project facts and user-provided assumptions, while also prohibiting the generation of offensive exploitation instructions.

security-testing