review-skills
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a bash script to discover installed skills by scanning directories such as
~/.agents/skills,~/.claude/skills, and project-local paths. This shell execution is limited to file discovery (find) and line counting (wc) to verify the presence of skill files before auditing them. - [INDIRECT_PROMPT_INJECTION]: The skill acts as an auditor for external
SKILL.mdfiles, which introduces a potential attack surface where instructions inside the audited skills could attempt to influence the agent's reporting behavior. - Ingestion points: Reads
SKILL.mdfiles and associated reference files from local and global agent directories. - Boundary markers: The instructions do not define specific delimiters or "ignore" directives to isolate the content being audited from the agent's execution context.
- Capability inventory: Perform shell-based file discovery, read local files, and generate a structured audit report.
- Sanitization: The skill relies on the agent's internal analysis against a fixed quality rubric rather than automated sanitization of the audited text.
Audit Metadata