The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external structured data which constitutes an attack surface for indirect prompt injection.\n- Ingestion points: The run.py script reads JSON data from a user-provided file path via the --input argument.\n- Boundary markers: There are no explicit instructions or delimiters in the SKILL.md or run.py to prevent the agent from following instructions embedded within the input data.\n- Capability inventory: The skill has filesystem access to read from and write to paths specified in command-line arguments.\n- Sanitization: The input is parsed as JSON but otherwise not sanitized or validated for malicious content before being processed.

security-usage-analyzer