ai-engineering-toolkit

Pass

Audited by Gen Agent Trust Hub on Aug 2, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill documentation references an external GitHub repository (github.com/viliawang-pm/ai-engineering-toolkit) and provides manual installation instructions using 'git clone' and a custom '/skill install' command.
  • [COMMAND_EXECUTION]: The README section includes examples of shell commands for manual setup (git clone, cp), but these are instructional and not automated by the skill during runtime.
  • [PROMPT_INJECTION]: The 'Agent Safety Guard' module is designed to generate adversarial test strings (e.g., jailbreak attempts, role-play bypasses) for the explicit purpose of security red-teaming and authorized evaluation of LLM applications.
  • [SAFE]: The skill functions as a decision-support and analysis tool; all findings are advisory and restricted to the evaluation context, with no evidence of unauthorized data exfiltration or credential harvesting.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 2, 2026, 05:41 PM
Security Audit — agent-trust-hub — ai-engineering-toolkit