The Agent Skills Directory

[SAFE]: The skill is a legitimate developer utility for plugin validation and scaffolding. The provided Python script (scripts/audit.py) and Jinja2 templates are transparent and perform standard development and linting tasks.
[COMMAND_EXECUTION]: The skill utilizes standard system tools (Bash, Python) for directory traversal, file reading, and scaffolding, which is appropriate and necessary for its stated purpose as an auditing and development tool.
[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill includes a surface for processing untrusted content from external plugin directories during the audit phase. While the logic is limited to structural validation and length checks, agents using this skill should be aware of the data ingestion point.
Ingestion points: The scripts/audit.py script reads the contents of SKILL.md and README.md files from a user-provided target directory.
Boundary markers: None. The script reads file contents into memory for basic validation checks (e.g., line count, character presence).
Capability inventory: The skill has access to the Bash, Read, and Write tools.
Sanitization: The script performs basic structural validation but does not sanitize the text content of the files it reads, as it currently only uses the data for static checks.

audit-plugin