security-expert

Pass

Audited by Gen Agent Trust Hub on Apr 9, 2026

Risk Level: SAFENO_CODEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [NO_CODE]: The skill is composed entirely of markdown instructions and does not include any scripts, binaries, or configuration files that execute code.
  • [PROMPT_INJECTION]: The instructions define a specific role and constraints for the agent to follow when reviewing content. There are no attempts to bypass safety filters, extract system prompts, or disregard core instructions.
  • [DATA_EXFILTRATION]: The skill lacks any tools or commands for network communication or accessing sensitive local files like credentials or SSH keys.
  • [SAFE]: The functionality is restricted to natural language analysis of documentation provided by the user, with no capability to perform side effects on the host system or external services.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 9, 2026, 04:19 PM