red-teaming-llms-with-garak

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The Python script scripts/agent.py invokes the garak CLI tool using subprocess.run(). The command is passed as a list of arguments rather than a concatenated string, which effectively mitigates the risk of shell injection attacks.
  • [EXTERNAL_DOWNLOADS]: The instructions direct users to install the garak package from PyPI or directly from NVIDIA's official GitHub repository. These are reputable and well-known sources appropriate for the skill's purpose.
  • [PROMPT_INJECTION]: Static analysis flagged terms such as 'DAN' and 'jailbreak' in SKILL.md. However, these are descriptive names for the security probes used by the red-teaming tool and do not constitute an attempt to inject malicious instructions into the agent.
  • [DATA_EXPOSURE]: The skill documents the use of environment variables (e.g., OPENAI_API_KEY) for providing credentials to the scanner, which is a secure and standard practice for managing sensitive API keys.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 04:11 AM
Security Audit — agent-trust-hub — red-teaming-llms-with-garak