review-skill

Pass

Audited by Gen Agent Trust Hub on Jun 15, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill is implemented as a read-only evaluation tool. It includes a comprehensive rubric for safety and quality (references/rubric.md) and follows best practices for tool execution by scoping searches to specific local directories.
  • [PROMPT_INJECTION]: The skill presents an inherent surface for indirect prompt injection as it is designed to ingest and process untrusted skill definitions. However, given its primary purpose as an auditing tool and the absence of high-risk capabilities like network access or file mutation, the risk is minimal.
  • Ingestion points: The skill reads all files from target skill directories (Step 1).
  • Boundary markers: Content is passed directly to an evaluation sub-agent without specific delimiters or isolation instructions.
  • Capability inventory: Execution is limited to text evaluation via sub-agents and read-only file access.
  • Sanitization: No sanitization of the input skill content is performed before analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 15, 2026, 10:15 AM