Blender Mcp

Integrate Blender MCP (Model Context Protocol) allowing OpenClaw to control Blender for 3D modeling, scene creation, and manipulation. Use when you need to i...

This skill is internally coherent for controlling a local Blender instance, but it grants high-power actions (notably 'execute arbitrary Python code' inside Blender) and references external asset/model services without declaring API credentials. Before installing: 1) Verify you trust the blender-mcp addon source (the README points to a GitHub repo) and the uvx package that will be invoked. 2) Inspect any addon.py you install and review the blender-mcp implementation for unexpected network behavior. 3) Use the included scripts (scripts/test-connection.py) in a controlled environment to confirm behavior. 4) If you use external downloads or model-generators, confirm whether API keys are needed and where they'll be stored. 5) Consider running Blender and the skill in an isolated VM/container or limit the agent’s autonomous invocation if you don’t want the agent to run potentially destructive Python inside Blender without oversight.

Type: OpenClaw Skill Name: openclaw-blender-mcp-skill Version: 0.1.0 The skill bundle provides integration with Blender via the Model Context Protocol (MCP) but includes high-risk capabilities, most notably an `execute_code` tool defined in `SKILL.md` that allows the AI agent to run arbitrary Python code within the Blender environment. Additionally, the skill includes tools for downloading assets from external sources (Poly Haven, Sketchfab) and a script (`scripts/start-server.sh`) that executes an external package via `uvx blender-mcp`. While these features are aligned with the stated goal of 3D modeling, the provision of arbitrary code execution and external network fetching represents a significant attack surface without built-in sanitization.