ai-engineering
Pass
Audited by Gen Agent Trust Hub on May 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides detailed architectural guidance for building AI systems. It includes a dedicated safety subskill (
subskills/safety-evals.md) that outlines principles for preventing prompt injection and securing tool interfaces. - [EXTERNAL_DOWNLOADS]: Documentation references target well-known and trusted technology companies and service providers, including OpenAI, Anthropic, Google, Microsoft, and MLflow. These are documented neutrally and used to support the provided engineering patterns.
- [COMMAND_EXECUTION]: A utility script
evals/run-trigger-evals.mjsis provided for testing purposes. It is a benign Node.js script that performs simple string classification and file I/O within the skill's own directory to verify its routing logic.
Audit Metadata