rmode
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to handle external inputs such as user code, technical specifications, and debugging requests. It proactively addresses this attack surface by instructing the agent to apply 'boundary-discipline' for parsing external input and explicitly requiring adherence to host security boundaries and task scope.
- [COMMAND_EXECUTION]: The skill directs the agent to perform verification ('verify-the-outcome') and create scripts or codemods to automate repetitive tasks ('build-the-lever'). These capabilities are constrained by instructions to follow existing repository gates, authorization policies, and security boundaries.
- [SAFE]: The skill references local sibling files (e.g.,
../principle-X/SKILL.md) to fetch additional instructions. These are references to the local file system within the expected skill directory structure and do not involve remote downloads or untrusted sources.
Audit Metadata