class-diagram
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests structural information from a codebase via the codebase-explorer agent to generate diagrams. This creates a potential vector for indirect prompt injection if a codebase contains malicious naming or metadata designed to influence the agent's behavior.
- Ingestion points: The skill receives structural models including class names, methods, and attributes from the cartograph:codebase-explorer agent (Step 1).
- Boundary markers: There are no explicit delimiters or instructions to ignore embedded instructions within the ingested codebase data.
- Capability inventory: The skill utilizes the Mermaid Chart MCP tool for rendering and the cartograph:codebase-explorer agent for data gathering. It does not have access to sensitive file system writes or unrestricted network operations.
- Sanitization: The skill transforms extracted data into Mermaid syntax but does not explicitly sanitize or validate these strings against potential injection patterns.
Audit Metadata