llm-top-10

Pass

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a static reference for security review procedures. It does not ship with executable scripts, binaries, or automated deployment logic.
  • [PROMPT_INJECTION]: The skill contains phrases typically associated with prompt injection (e.g., 'ignore your previous instructions', 'reveal your system prompt', 'DAN'). Analysis shows these are present in two safe contexts: 1) within the descriptions of the OWASP LLM01 and LLM07 categories to help the agent identify vulnerabilities in the code it is reviewing, and 2) in Section 8 (Prompt Injection Safety Notice), which explicitly warns the agent to treat such strings in analyzed content as data rather than instructions.
  • [COMMAND_EXECUTION]: The skill restricts itself to read-only tools (Read, Grep, Glob) via the frontmatter configuration. It contains no shell commands or subprocess calls that would execute user-controlled data.
  • [DATA_EXFILTRATION]: There are no network-capable tools or outbound requests. References to sensitive files (e.g., .env, credentials) are used to instruct the agent on what sensitive information to look for during a manual code review.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 20, 2026, 02:11 PM
Security Audit — agent-trust-hub — llm-top-10