ai-testing-safety

Pass

Audited by Gen Agent Trust Hub on May 13, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill files (SKILL.md, examples.md, and templates/red_team_suite.py) contain a high volume of strings that are typical prompt injection and jailbreak payloads (e.g., "ignore all previous instructions", "reveal the system prompt", "Override safety"). These are correctly utilized as test data (target behaviors) for the adversarial auditing of other systems, rather than as instructions for the AI agent itself.
  • [EXTERNAL_DOWNLOADS]: The documentation suggests installing a related skill from the same author using the command npx skills add lebsral/DSPy-Programming-not-prompting-LMs-skills. This is a vendor-owned resource used to extend functionality and is considered a safe operational practice.
  • [COMMAND_EXECUTION]: The skill provides templates for defining and executing Python functions (target_system) that encapsulate the AI models being tested. This facilitates the intended core functionality of sending generated attack prompts to a target system and evaluating the output for vulnerabilities.
Audit Metadata
Risk Level
SAFE
Analyzed
May 13, 2026, 06:46 PM
Security Audit — agent-trust-hub — ai-testing-safety