skill-auditor
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a defensive security tool designed to guide an agent through a structured vetting process for other agent skills.
- [INDIRECT_PROMPT_INJECTION]: As an auditor, the skill is intended to process untrusted skill files. This creates an attack surface for indirect prompt injection; however, the skill explicitly provides instructions and patterns to detect and mitigate these risks (Step 4). The skill restricts itself to file-read permissions, significantly limiting the potential impact of a malicious payload. 1. Ingestion points: SKILL.md body and frontmatter of skills being audited. 2. Boundary markers: None specified. 3. Capability inventory: file-read (SKILL.md). 4. Sanitization: The skill performs text normalization (Base64 decoding, Unicode expansion) as part of its scanning protocol.
- [PROMPT_INJECTION]: Static analysis identified strings such as 'Ignore previous instructions' and 'DAN/jailbreak'. These are present within a defensive context as examples of patterns to flag during an audit and do not constitute an attempt to subvert the agent's behavior.
Audit Metadata