reflect
Pass
Audited by Gen Agent Trust Hub on Jul 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security threats detected. The skill instructions are limited to internal reasoning and updating local project-specific context files such as AGENTS.md and CONTEXT.md, which is consistent with standard agentic workflows for maintaining state.
- [SAFE]: The skill metadata includes 'disable-model-invocation: true', which is a restrictive configuration that prevents the agent from calling additional tools while executing these instructions, significantly reducing the attack surface.
Audit Metadata