skill-reviewer

Pass

Audited by Gen Agent Trust Hub on Jul 3, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill is designed to process external, potentially untrusted content in the form of other SKILL.md files for review.
  • Ingestion points: User-provided skill text or local file paths referenced in SKILL.md (Step 1).
  • Boundary markers: The instructions do not define clear delimiters or "ignore" instructions to prevent the agent from following commands embedded within the skill being reviewed.
  • Capability inventory: Across SKILL.md, the skill is limited to generating text-based review reports; it does not contain tools for subprocess execution, network activity, or writing to the filesystem.
  • Sanitization: There are no mechanisms described for sanitizing or validating the input data before it is processed by the agent.
  • [DATA_EXFILTRATION]: Step 1 of SKILL.md instructs the agent to read from a user-provided file path. While intended for reviewing skill files in a repository, this capability could be exploited by a malicious user to prompt the agent to read and reveal the contents of sensitive files (e.g., credentials or configuration files) if the agent's host environment permits such access.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 3, 2026, 01:10 PM
Security Audit — agent-trust-hub — skill-reviewer