d2
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied natural language instructions to generate D2 diagram code, which is then passed to external tools for inspection and rendering.
- Ingestion points: User requests for architecture diagrams, flowcharts, and sequence diagrams, as well as contents of existing .d2 files.
- Boundary markers: The instructions do not define specific delimiters or "ignore" instructions to prevent the agent from following commands embedded within user-provided diagram labels or descriptions.
- Capability inventory: The skill utilizes
d2_inspect,d2_render,d2_validate, and file writing tools to process the generated code. - Sanitization: There is no evidence of sanitization or escaping mechanisms to prevent user-provided strings from influencing the final D2 code logic beyond its intended structure.
- [EXTERNAL_DOWNLOADS]: The skill instructions include patterns for incorporating SVG icons from
https://icons.terrastruct.com. This is a well-known service associated with the D2 diagramming language for providing technical icons (AWS, GCP, React, etc.).
Audit Metadata