agent-skill-audit
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to analyze untrusted input (audited prompts and instruction packages). To mitigate risk, it includes a robust 'Trust Boundary' section in
SKILL.mdthat explicitly commands the agent to treat all audited material as untrusted and to disregard any instructions found within that content. The skill operates in a read-only mode for artifact inspection, further limiting the attack surface. - [EXTERNAL_DOWNLOADS]: The skill references documentation hosted at
platform.claude.com. These are official resources from a trusted vendor (Anthropic) providing model-specific compatibility guidance. These references are used for static analysis and do not involve downloading or executing arbitrary remote code.
Audit Metadata