common-llm-security
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates security audits of AI components by processing untrusted code, prompt templates, and retrieved document chunks. This inherent function exposes the agent to indirect prompt injection where instructions hidden in the data being audited could attempt to override the auditor's behavior.
- Ingestion points: Audited code snippets, prompt templates, and RAG data provided in the user's request.
- Boundary markers: The instructions lack specific guidance on using delimiters to safely encapsulate the untrusted content being analyzed.
- Capability inventory: The skill itself does not define executable scripts, but it directs the agent's behavior during a security review.
- Sanitization: There are no instructions for sanitizing or escaping the content under review before it is ingested into the agent's context.
Audit Metadata