Sandbox
@sandraschi/blender-mcp

MCP server for Blender automation

Blender MCP connects an agent to Blender so it can build scenes, run mesh and animation operations, export assets, and render output. It runs Blender headless by default, and can switch to a live viewport when the bridge addon is installed. The repo also ships a dashboard and a Tauri desktop app for managing the workflow and watching the scene pipeline.

43 stars10 forksPythonUpdated 11d ago
Who it's for

Builders who want Cursor or another MCP client to drive Blender scenes, exports, and renders.

What it delivers

You can create and export Blender work from agent prompts instead of opening Blender for every step.

What it does

Headless Blender execution

Runs `blender --background` for batch mesh, VSE, Grease Pencil, export, and render jobs.

Live viewport bridge

Uses `addon/blender_bridge_addon.py` to connect a running Blender session when you want to watch updates live.

MCP tool set

Exposes dozens of Blender actions through FastMCP, including scene edits, animation, compositing, and export flows.

VRM and asset pipelines

Includes export and validation paths for VRM avatars, GLB/GLTF assets, and Gaussian splats.

Dashboard and desktop app

Provides a web dashboard and a Tauri app for scene exploration and pipeline control.

README

Blender MCP — AI-Powered Blender Automation

Control Blender with natural language through MCP. Tell Claude to create a steampunk robot with glowing eyes and watch it build in Blender.

Stars License Python FastMCP

Get started in 2 minutes

DownloadWhat it is
blender-mcp-*.mcpbMCP server for Claude Desktop
Blender3D app (install separately, never bundled)
blender_bridge_addon.py (optional)Live viewport only — skip for headless use

Steps

  1. Download the .mcpb from Releases
  2. Drag it into Claude Desktop (Settings → MCP Servers → Install from file)
  3. Install Blender if you have not already
  4. Restart Claude Desktop and try: Create a red cube on a gray floor with a sun lamp.

That is enough for most tools — the server runs Blender headless in the background.

Optional — live viewport: install the bridge addon in Blender to watch the scene update while you chat. See INSTALL.md.

Other install methods (npx mcpb, manual config, developer setup): INSTALL.md

How it runs

ModeHost appWhen
Headless (default)blender --background subprocessBatch export, CI, agents without a display; VSE, geonodes, most mesh ops
Live GUI (optional)Blender + bridge addonWatch the agent build; viewport screenshots; sculpt with live feedback
Per-tool overrideprefer_session=FalseBatch jobs force headless even if bridge is connected

You do not need to open Blender's UI for most MCP tools — the server spawns headless Blender automatically and falls back from live session when no bridge is connected.

Install Blender separately; it is never bundled. Override path with BLENDER_EXECUTABLE.

Watch or batch — Most tools run headless. Use blender_session start + the bridge addon only if you want the viewport to update live. See INSTALL.md.

Hands-in / Hands-out

DirectionArtifactsNotes
Hands-inNatural-language scene promptsAgent instructions; blender_ai_* script generation
Hands-in.blend, image refs, mesh filesWebapp upload or tool params
Hands-inRodin / Tripo / Hunyuan mesh URLsblender_ai_generate and related tools
Hands-inInline bpy scriptsblender_script_execute, handler-backed tools
Hands-out.glb, .gltf, .fbx, .obj, .usdblender_exportheadless
Hands-out.vrm, VRChat-ready avatarsExport + validation pipeline — headless
Hands-out.blend (saved scene)After agent edit session — headless or live bridge
Hands-outViewport PNG, MP4 (VSE), Gaussian splatsblender_render, blender_vse, splat tools — headless

Fleet pipelines (downstream)

Downstream MCPTakes from blender-mcp
godot-mcp.glb / .gltf game assets
vrchat-mcp.vrm after validation
tahoma2d-mcpRendered image sequences / GP output
freecad-mcp.step via intermediate export
unity3d-mcp.fbx / .glb for Unity import

Features

  • Natural-language 3D creation — scenes, meshes, materials, lighting, animation
  • Live GUI bridge — watch the agent build in Blender while you chat
  • 48+ MCP tools — mesh edit, sculpt, geonodes, compositor, VSE, export (GLB, VRM, VRChat, Unity)
  • Generative AI hooks — Rodin/Tripo/Hunyuan mesh generation and vision refine
  • Webapp dashboard — scene explorer, agent lab, materials, mesh/splat pipeline
  • Fleet-ready — FastMCP 3.2, .mcpb packaging, Prometheus metrics, optional Docker

What You Can Do

Try these in Claude Desktop after install:

Create a red cube on a gray floor with a sun lamp and render a viewport screenshot.

Build a simple chair with wood material and export as GLB.

Start a live Blender session and add a sphere with a metallic blue shader.

Documentation

DocContents
InstallationAll install methods, prerequisites, bridge setup
ConfigurationEnv vars, Claude Desktop env block
Tool ReferenceCapabilities and tool catalog
DevelopmentContributing, just, lint, build
TroubleshootingCommon errors and fixes
ArchitectureSystem design
RoadmapPlanned improvements
MonitoringPrometheus / Grafana / Loki
DockerOptional container deploy

Extended guides: docs/DOCUMENTATION_INDEX.md

Webapp and Native App

Dashboard (optional): .\start.ps1http://localhost:10848 — see INSTALL.md.

Tauri desktop installer (~15 MB, no Python required): just build-native — see docs/DEVELOPMENT.md.

Requirements

  • Claude Desktop (or any MCP client for manual config)
  • Blender 3.0+ — installed separately, never bundled; override with BLENDER_EXECUTABLE
  • LLM (optional) — Ollama, LM Studio, or cloud API for script generation; see INSTALL.md
  • OS: Windows, macOS, Linux
  • Python 3.12+ — only for Options C/D (clone-from-source)

License

MIT — FlowEngineer sandraschi. Free for personal and commercial use.

Files in the repo

Repository payload73 top-level entries
  • _llm_test_scripts
  • .claude-plugin
  • .cursor
  • .github
  • .opencode
  • .snapshots
  • addon
  • assets
  • data
  • docs
  • docs-private
  • examples
  • hooks
  • mcpb
  • mcpb-packaging
  • monitoring
  • native
  • scripts
  • skills
  • src
  • src-tauri
  • systemd
  • test_data
  • tests
  • webapp
  • .cursorrules
  • .depot.json
  • .dockerignore
  • .env.example
  • .gitattributes
  • .gitignore
  • .mcpbignore
  • .pre-commit-config.yaml
  • .python-version
  • .windsurfrules
  • AGENTS.md
  • ASSESSMENT.md
  • blender-mcp-backend.spec
  • BUILD_LOG.md
  • build.ps1
  • Cargo.toml
  • CHANGELOG.md
  • check_result.py
  • claude_download_guide.txt
  • CLAUDE.md
  • docker-compose.yml
  • Dockerfile
  • extension.toml
  • extension.wasm
  • fleet-start.config.ps1
  • glama.json
  • INSTALL.md
  • justfile
  • LICENSE
  • llms-full.txt
  • llms.txt
  • Makefile
  • manifest.json
  • mcpb.json
  • pyproject.toml
  • README_NEW.md
  • README.md
  • requirements-dev.txt
  • restart.bat
  • run_server.py
  • setup.py
  • start.bat
  • start.ps1
  • STATUS.md
  • stop.bat
  • TODO.md
  • uv.lock
  • verify_tools.py

Discussion (0)

Ask about usage, or say what you built with it

Sign in to join the discussion.

No comments yet. Be the first to say what this is good for.

More connectors

High-performance code intelligence MCP server. Indexes codebases into a persistent knowledge graph — average repo in milliseconds. 158 languages, sub-ms queries, 99% fewer tokens. Single static binary, zero dependencies.

43k

Universal provider proxy for OpenAI Codex & Claude Code — use any LLM (Claude, Gemini, Grok, DeepSeek, Ollama…) with Codex CLI, App, SDK, and Claude Code

14k
okf-memory/
okf-agent-memory

Git-native persistent memory for AI coding agents. Implements Google OKF v0.2 with sub-300µs in-memory BM25 search, embedded MCP server, and progressive disclosure. Slashes token bloat by 80% with zero external databases or dependencies. Built in pure Go.

547
tirth8205/
code-review-graph

Local-first code intelligence graph for MCP and CLI. Builds a persistent map of your codebase so AI coding tools read only what matters, with benchmarked context reductions on reviews and large-repo workflows.

31k
2akouwu/
reverify

Stop your AI from making things up — it proposes, deterministic tools decide, every claim checked against ground truth with evidence. Grounded facts and context survive resets. Reverse engineering is the proving ground. MCP server + CLI.

1.1k
t8y2/dbxConnectors

20 MB lightweight cross-platform database client for 90+ databases, including MySQL, PostgreSQL, SQLite, Redis, MongoDB, DuckDB, SQL Server, and Dameng. Built-in AI, MCP Server, CLI, desktop and Docker. | 轻量级跨平台数据库管理工具,支持 MySQL、PostgreSQL、SQLite、Redis、MongoDB、达梦等 90+ 数据库,提供桌面端、Docker、CLI、内置 AI 助手和 MCP Server。

19k