The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from user-specified skill directories, which could contain malicious instructions designed to influence the auditing agent's behavior.
Ingestion points: Reads SKILL.md, README.md, and various files within the references/, scripts/, and workflows/ directories of the skill being audited.
Boundary markers: Absent. The instructions do not provide explicit delimiters or warnings to ignore potentially malicious instructions embedded within the files being analyzed.
Capability inventory: The skill utilizes WebFetch for network operations (URL verification) and performs extensive file reading across the provided directory.
Sanitization: Absent. The skill extracts and reports content directly from the analyzed files without escaping or validation to prevent instruction override.

skill-review