cadcraft-mcp

Pass

Audited by Gen Agent Trust Hub on Nov 4, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes cadcraft-cli and a launch.py script to perform CAD operations and manage sessions. It specifically mentions connecting to local services at 127.0.0.1:PORT to bridge between headless and desktop environments.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by processing external drawing data and entity attributes which are then returned to the agent context.
  • Ingestion points: Data retrieved via inspect_drawing, query_entities, and render functions.
  • Boundary markers: No specific delimiters or safety instructions are defined to separate untrusted CAD data from the agent's primary instructions.
  • Capability inventory: Includes file system write access (save), arbitrary CAD command execution (execute, command_line), and UI automation (ui_click).
  • Sanitization: No sanitization or validation of the ingested CAD data is described in the instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 4, 2026, 06:48 PM