cadcraft-use
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill acts as an instructional entry point for CAD operations, directing the agent to use either the MCP or CLI versions of the CADCraft toolset. All referenced repositories and skills originate from the same vendor (full-aigc-skills).
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a data ingestion surface by processing external CAD files (DXF, DWG, PDF).
- Ingestion points: External drawing files and their metadata (layers, entities) processed by the agent.
- Boundary markers: No explicit data delimiters are defined in this routing skill, though it provides instructions for rigorous verification of tool outputs.
- Capability inventory: The skill uses tools for inspecting drawings, querying entities, rendering content, and connecting to a local desktop application via a loopback address.
- Sanitization: Validation is performed through post-execution checks, such as re-opening files and verifying entity counts and types.
- [SAFE]: The instructions for connecting to the CADCraft desktop application via localhost (127.0.0.1) are consistent with standard automation practices for local software integration and do not involve unauthorized network exposure.
Audit Metadata