Blender MCP
Connects an MCP client to a running Blender session via a bridge add-on, exposing tools to check bridge/application status and scripting permission, retrieve scene context (units, object count, and up to 100 selected objects), and optionally execute user-confirmed native Python through Blender's API to inspect or modify the scene.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Blender MCPCheck my Blender connection and describe what's currently selected."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Blender MCP by ArchBench
Connect an MCP-compatible assistant to Blender. Read scene and selection context, then optionally run user-requested Python through Blender's native API.
Experimental alpha. Native acceptance tests passed on Blender 4.3.2 on macOS. Other Blender versions and interactive installation/viewport/Undo workflows remain unqualified. Use a disposable scene first. See validation.
The connector is independently implemented. It requires no ArchBench account, subscription, or desktop application and sends no telemetry.
Install
Prerequisites: Blender desktop, Node.js 22+, and Python 3 for building the add-on. The implementation targets Blender 3.6+, with the qualification limit above.
git clone https://github.com/3d-mcp/blender-mcp.git
cd blender-mcp
npm ci --ignore-scripts
npm run package:hostsA prebuilt add-on is provided on the alpha release page.
In Blender Preferences, install dist/blender-mcp-addon.zip (or the downloaded ZIP) using Add-ons →
Install from Disk (wording varies by version), then enable the add-on. In the
3D View sidebar, open MCP and click Start MCP bridge. Leave Allow native
scripts unchecked for read-only use. This is a legacy add-on distributed through
GitHub, not a listing in Blender's official Extensions catalog.
Configure your MCP client using absolute paths, for example:
{
"mcpServers": {
"blender": {
"command": "/absolute/path/to/node",
"args": ["/absolute/path/to/blender-mcp/bin/blender-mcp.mjs"]
}
}
}Client setup and troubleshooting. The client launches the stdio process; no separate server terminal is needed. Both processes must run on the same machine under the same OS user. There is no npm registry release yet.
Related MCP server: Blender MCP
First request
“Check my Blender connection and describe the current selection.”
Tool | Behavior |
| Bridge/application versions and scripting permission |
| Scene, units, object count, and up to 100 selected objects |
| Optional, explicitly enabled native Python |
To use scripting, stop the bridge, check Allow native scripts, and restart.
Also set ARCHBENCH_MCP_ALLOW_SCRIPTS=1 in the MCP client's server environment.
Each execution requires confirm: true. Example Python:
result = {"objects": len(bpy.context.scene.objects)}Scripts have Blender's full authority, including file and network access. Return
small JSON-compatible values in result. Timeout/cancellation can prevent queued
work from starting but cannot interrupt a running script or undo its effects.
Never automatically retry a mutation. Security and limits.
Development
npm run check
node scripts/test-blender-native.mjs /path/to/BlenderThe first command runs MCP/transport regressions, source syntax and dependency checks, and verifies the packaged source and licenses. The second runs native acceptance tests in an isolated, unsaved scene. GitHub Actions runs the transport checks on Linux, macOS and Windows; those checks do not certify native Blender on all three operating systems.
License and ArchBench
The independent MCP server and shared transport are MIT. The Blender add-on is GPL-3.0-or-later; its archive includes all corresponding Python source and both license notices. See LICENSING.md for the exact file mapping.
Maintained by ArchBench, the separate product for AI-assisted design work. ArchBench's private modeling intelligence, project history, collaboration and workflow systems are not included here. The connector remains usable independently.
This is an independent integration, not an official Blender project.
This server cannot be deployed
Maintenance
Related MCP Connectors
- ZapierOAuthcom.zapier
Hosted MCP server connecting AI assistants to 9,000+ apps and 40,000+ actions via Zapier.
Connect AI agents to Replynodes over the Model Context Protocol.
Cloud Blender for AI agents: scenes, assets, renders, MP4, STL, GLB — over hosted remote MCP.
Personal assistant MCP server with search, execute, packages, jobs, secrets, and integrations.
Related MCP Servers
- AlicenseBqualityCmaintenanceConnects Blender to AI through the Model Context Protocol to enable prompt-assisted 3D modeling and scene manipulation. It supports object creation, material control, and arbitrary Python code execution directly within the Blender environment.22MIT
- AlicenseNot gradedqualityCmaintenanceConnects Blender to AI assistants through the Model Context Protocol, enabling direct AI control of 3D modeling, scene creation, and manipulation via natural language.MIT
- AlicenseNot gradedqualityBmaintenanceConnects Blender to local LLMs through MCP, enabling AI-assisted 3D modeling, scene creation, and manipulation.1MIT
- AlicenseNot gradedqualityCmaintenanceConnects MCP-compatible clients to a live Blender scene for AI-assisted 3D workflows, enabling inspection and controlled operations on objects, materials, cameras, lights, render settings, animation, UVs, Geometry Nodes, imports, exports, and Python execution.1MIT