nooa-agent-authoring
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a documentation-focused instruction set for a legitimate development framework, providing implementation details rather than executing dangerous actions.
- [PROMPT_INJECTION]: The documentation explicitly warns against unsafe prompt construction practices, such as interpolating untrusted data directly into instructions, and recommends using structured strategies to maintain a clean instruction channel.
- [CREDENTIALS_UNSAFE]: The skill demonstrates secure secret management by showing how to use the
@hiddendecorator for sensitive fields like API keys, ensuring they are not exposed to the LLM's observation space. - [DYNAMIC_EXECUTION]: The framework features a
CodeActStrategyfor LLM-driven code execution. This is a documented architectural pattern for agent development that includes safety controls such as iteration limits, validation retries, and cell timeouts.
Audit Metadata