The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied file or directory paths via the $ARGUMENTS variable to compare them against best practice guidelines. This ingestion of external data constitutes a surface for indirect prompt injection. * Ingestion points: The $ARGUMENTS placeholder in SKILL.md captures user input for targeted examination. * Boundary markers: User input is delimited by <user_input> tags. * Capability inventory: The skill is restricted to providing guidance and references; no active capabilities for system modification, network access, or command execution are present in this file. * Sanitization: There is no evidence of sanitization or content filtering for the data residing in the user-specified paths.

typer-and-rich