diagrams
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists of instructional content for documentation best practices. It does not contain any executable scripts, privilege escalation attempts, or unauthorized network operations.
- [INDIRECT_PROMPT_INJECTION]: The skill recognizes that data processed for diagrams (like labels and datasets) could contain malicious instructions. It provides specific guidance to mitigate this surface.
- Ingestion points: Processes labels, URLs, imported datasets, and issue text during diagram modeling (SKILL.md).
- Boundary markers: Instructions require establishing a clear contract for the diagram's question, evidence, and destination.
- Capability inventory: The skill is strictly instructional and does not define any command execution or network capabilities for the agent to perform on its own.
- Sanitization: The instructions explicitly direct the agent to treat evidence as untrusted, avoid executing embedded instructions, and not enable unsafe HTML, even if the renderer supports it.
Audit Metadata