The Agent Skills Directory

Indirect Prompt Injection (HIGH): The api_validator.py script creates an attack surface by processing untrusted data from a user-specified path.
Ingestion points: The script reads file names and file content from the <project_path> provided as a command-line argument.
Boundary markers: The script does not use any delimiters or safety instructions to distinguish between its own reporting and the data being analyzed, which could allow the agent to mistake file content for instructions.
Capability inventory: The agent possesses Write and Edit tools, meaning any malicious instructions discovered and displayed by the validator could potentially influence the agent to perform unauthorized file changes.
Sanitization: There is no sanitization or escaping of the content read from the external project files before it is output to the agent's context.

api-patterns