blender-ai-mcp

Overview

blender-ai-mcp provides a stable tool API for controlling Blender through MCP clients instead of generating raw Python scripts. It implements a production approach with atomic tools, macro tools, and workflow tools, offering goal-first orchestration and deterministic verification. The server features a small public surface with search-first bootstrap layer and supports multiple vision runtimes including local MLX and external providers.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• router_set_goal

  Set the modeling goal for the current session

  Try:Set goal: create a detailed character modelTry:Goal: design a modern chair
• router_get_status

  Get the current status of the router session

  Try:Check session statusTry:What is the current goal status?
• browse_workflows

  Browse available modeling workflows

  Try:Show me available workflowsTry:List character modeling workflows
• search_tools

  Search for tools based on criteria

  Try:Find tools for mesh editingTry:Search for lighting tools
• call_tool

  Execute a specific tool with arguments

  Try:Call tool to create a primitive cubeTry:Use the merge vertices tool
• check_scene

  Check the current scene state

  Try:What objects are in the scene?Try:Check scene for issues
• inspect_scene

  Inspect specific objects in the scene

  Try:Inspect the main character objectTry:Check properties of the camera
• configure_scene

  Configure scene settings

  Try:Set scene render engineTry:Configure scene units
• macro_cutout_recess

  Create recesses, openings, and cutter-driven cutouts

  Try:Create a door recess in the wallTry:Add window openings
• reference_images

  Manage reference images for modeling goals

  Try:Attach reference images for the characterTry:Update reference images
• scene_scope_graph

  Get structural scope graph of the scene

  Try:Show scene structure graphTry:Get object hierarchy relations
• scene_view_diagnostics

  Get view diagnostics for scene cameras

  Try:Check camera view coverageTry:Analyze viewport visibility

Comparable tools

Installation

pip install blender-ai-mcp

{
  "mcpServers": {
    "blender-ai-mcp": {
      "command": "python",
      "args": ["-m", "blender_ai_mcp.server"]
    }
  }
}

FAQ

How does this differ from raw Python scripting?

This provides stable tool contracts with validated parameters instead of improvising Blender code, offering better safety, reliability, and deterministic verification.

What vision providers are supported?

Local MLX plus external OpenRouter and Google AI Studio / Gemini provider paths with model-family-specific contract profiles.

Compare blender-ai-mcp with