TrendWhat rose on GitHub, every day at 07:30 KST · 한국어

GitHub · as of November 15, 2026 · View on GitHub →

ahujasid/mcp-for-blender

Blender MCP server

+863 stars in 7 days

Connect AI to Blender for automated 3D modeling and scene composition

When given a prompt, the AI creates objects, applies materials, and verifies the results within Blender. It works by writing and executing Blender Python code directly, then using visual feedback to improve accuracy. This allows repetitive modeling tasks and complex scene setups, previously done manually, to be handled interactively.

Renaming and Legacy Environment Compatibility

The repository was previously known as blender-mcp, but the current PyPI package name is mcp-for-blender. Existing environments continue to work without any configuration changes, and the uvx blender-mcp command still runs the server. For new installations, using mcp-for-blender is recommended. This change aims for consistency in package management and does not affect existing user workflows.

Scope of 3D Tasks Performed by AI

The AI writes Blender Python code directly to perform modeling, layout, material application, animation, rigging, and retopology tasks.

  • Visual Verification The AI checks its own work using various methods such as camera views, wireframes, X-ray, and rendering to identify and correct errors.
  • Asset Integration It can search for CC0 licensed assets from Poly Haven and models from Sketchfab and Poly Pizza, applying them directly to the scene.
  • Model Generation 3D models are generated via text prompts or reference images using Meshy, Tripo, Hunyuan3D, and Hyper3D Rodin.

Installation and Connection Process

Installation is based on the uv package manager, with automatic installation scripts available via curl on macOS and Linux, and PowerShell on Windows. The script installs uv, detects the user's MCP client (such as Claude Desktop, Cursor, or VS Code) to modify configuration files, and installs and activates the Blender add-on. Existing configuration files are backed up, and only the blender entry is added. For manual installation, install uv, specify uvx mcp-for-blender as the server command in the MCP client settings, install the add-on with uvx mcp-for-blender install-addon, and then activate it in Blender. When Blender opens, the add-on starts a socket server and connects to the AI client.

Security and Safe Mode

By default, the AI can execute arbitrary Python code in Blender. Setting the BLENDER_MCP_SAFE_MODE=1 environment variable ensures scripts are validated before execution, blocking dangerous actions such as file input/output, external program execution, network access, and persistent code. Blocked scripts are returned to the AI with the reason, allowing for a corrected retry. Normal Blender operations (modeling, materials, rendering, saving, exporting) remain possible in safe mode. Since the add-on's socket server provides no authentication or encryption, anyone with access to the port can execute Python in Blender. Therefore, restricting access to localhost or using an SSH tunnel is recommended on untrusted networks.

Pre-Application Checks

This tool requires Blender 3.0 or later, Python 3.10 or later, and the uv package manager. MCP clients launched from a GUI do not inherit the terminal's PATH, so the full path to uvx may need to be specified in the settings. In environments using Python version managers like conda, pyenv, or asdf, uv might select an unexpected interpreter; it is advisable to pin the Python version using --python 3.11 and UV_PYTHON_PREFERENCE=only-managed. When running the MCP server with Docker, Blender runs on the host machine while the container hosts only the server. On Linux, host.docker.internal does not exist, so host networking mode must be used. This project is a third-party integration unrelated to the Blender Foundation, and users are responsible for the results of AI-generated code execution.

Written by AI from this repository's README on October 29, 2026. For the facts, the original on GitHub is the reference.

By the numbers

GitHub trending history
1 day between Jun 19, when Trend began recording, and Nov 15, 2026 · first on Sep 29

Read from GitHub, as of November 15, 2026.

Language
Python
Topics
3d-modeling · ai · blender · blender-addon · blender-mcp · claude · generative-ai · llm
Stars
34,513
Stars gained in 7 days
+863
Contributors
47
Forks
3,151
Commits
273
Last commit
November 15, 2026
Open issues
28
Open pull requests
31

Related repositories

View on GitHub → · Homepage