blender-mcp

What it is and what it does

BlenderMCP bridges Claude AI and Blender through the Model Context Protocol, creating a two-way communication channel that lets Claude read scene data, create and modify 3D objects, apply materials, and execute Python code directly inside Blender. You install a lightweight Python server (via uvx or pipx) that connects to a Blender addon, then configure it in Claude Desktop, Cursor, or VS Code. Once set up, you can describe what you want in natural language and Claude will translate those instructions into Blender operations—creating objects, adjusting materials, inspecting the scene, or running custom Python scripts.

The package depends on mcp and httpx for its core networking and protocol handling. Installation is straightforward but requires upfront setup: installing the uv package manager, configuring your MCP client, and adding the Blender addon. The project is actively maintained (latest release 11 days old) and carries no known security vulnerabilities. It's designed for users who want AI-assisted 3D modeling workflows, whether for rapid prototyping, scene generation, or automating repetitive Blender tasks through conversational prompts.

Use it for:

• Rapidly prototype 3D scenes by describing them to Claude, which then creates and arranges objects in Blender without manual clicking.
• Generate 3D models from text descriptions using integrated services like Hunyuan3D or Hyper3D Rodin, controlled through Claude's prompts.
• Search and download assets from Sketchfab or Poly Haven directly into your Blender scene via Claude's natural language commands.
• Automate repetitive Blender tasks—batch object creation, material application, or scene inspection—by writing prompts instead of manual scripts.
• Debug or inspect complex Blender scenes by asking Claude to query scene data and report back on object properties, materials, and structure.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

BlenderMCP is a Model Context Protocol server that enables Claude AI to directly control and interact with Blender, allowing AI-assisted 3D modeling, scene creation, and object manipulation through natural language prompts.

Yes, if you use Blender and Claude Desktop (or Cursor/VS Code with MCP support) and want to experiment with AI-assisted 3D modeling. The package is actively maintained, has no security vulnerabilities, and carries a permissive MIT license. Installation requires upfront setup (uv, Blender addon, client configuration) but is well-documented. Start with it if you're curious about AI-driven workflows in 3D; it's low-risk to try and uninstall.

Install

blender-mcp on PyPI

pip install blender-mcp

uv add blender-mcp

poetry add blender-mcp

Installing blender-mcp

Before you install

Low friction installation via uvx or pipx; package is actively maintained with a recent release (11 days old) and depends only on mcp and httpx, both lightweight runtime dependencies.

License in practice

MIT license is permissive; you can use, modify, and distribute this package freely with minimal restrictions, making it suitable for both personal and commercial projects.

Quickstart

# Install via uv (recommended)
uvx blender-mcp

# Or via pipx
pipx install blender-mcp

# Configure in Claude Desktop's claude_desktop_config.json:
# {
#   "mcpServers": {
#     "blender": {
#       "command": "uvx",
#       "args": ["blender-mcp"]
#     }
#   }
# }

# Then install the Blender addon (addon.py) in Blender:
# Edit > Preferences > Add-ons > Install > select addon.py

Requires Blender 3.0 or newer, Python 3.10 or newer, and the uv package manager (or pipx as fallback). The Blender addon must be installed separately and enabled in Blender's preferences. MCP clients launched from a GUI may not inherit terminal PATH, requiring the full path to uvx.

Verify before relying

• Whether the package works with Blender versions between 3.0 and the current release without known compatibility issues.
• Performance characteristics when controlling complex scenes or running large batches of commands through Claude.
• Whether remote Blender connections (via BLENDER_HOST environment variable) are production-ready or experimental.

Package facts

Evidence: blender_mcp-1.8.0-py3-none-any.whl

Tags

More Artificial Intelligence packages

Further reading