blender-mcp-ultra
Provides comprehensive MCP tools for controlling Blender, including modeling, coloring, animation, rigging, geometry nodes, UV mapping, rendering, and scene management.
Allows the internal Blender chat assistant to use Google models for natural language command processing.
Allows the internal Blender chat assistant to use locally hosted Ollama models for natural language command processing.
Allows the internal Blender chat assistant to use OpenAI models for natural language command processing.
Click on "Install 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-mcp-ultraModel a low-poly tree and add a wind animation"
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-ultra
MCP server lengkap untuk Blender — 138+ tool, aman, teruji di Blender 4.2 LTS nyata.
Kendalikan Blender dari asisten AI mana pun (Claude, Cursor, Windsurf, opencode)
lewat Model Context Protocol (MCP), atau dari chat internal Blender dengan LLM
OpenAI-compatible. Seluruh permukaan tool dibangun dari satu spesifikasi
(mcp_tools.py) yang melayani FastMCP (SSE) dan bridge STDIO sekaligus.
Demo End-to-End (LLM Nyata → MCP → Blender Headless)
Adegan di bawah dibangun oleh model gpt-5.6-sol melalui API OpenAI-compatible:
model memilih 34 aksi tool MCP, dieksekusi lewat socket TCP :9876 ke
Blender 4.2.23 LTS blender -b (headless), lalu dirender dengan Cycles.
Frame 1 — meja kayu, vas kaca, bola logam, kursi hijau, lantai | Frame 60 — bola teranimasi (lokasi + rotasi) | Sudut kamera kedua |
|
|
|
Yang dikerjakan LLM dalam demo:
Modeling —
create_object(CUBE/CYLINDER/SPHERE/GRID) +transform_objectColoring —
create_material×5 (kayu, kacatransmission+ior, logammetallic, lantai, hijau) +assign_material×5Animation —
animate_location+animate_rotation(bola, frame 1..120)Rigging —
create_armature,rig_from_scratch(kursi),add_armature_modifierScene/Render —
create_camera,create_light(AREA 300 W),set_render_engine(CYCLES),render_frame
Detail lengkap: DEMO_LLM.md. Jalur verifikasi penuh
(70/70 + 66/66 perintah di Blender nyata, socket E2E headless) ada di sana.
Related MCP server: Blender MCP for Antigravity
Fitur
138+ tool MCP satu spesifikasi (
mcp_tools.py) → FastMCP + STDIOAlur "dari nol" per domain:
model_from_scratch,colorize_from_scratch,rig_from_scratch,animate_from_scratchModeling data-API aman di
blender -b: 12 primitive, 25+ modifier, bmesh (bevel/extrude/inset/loop cut/merge), kurva, teks 3D, screw/latheColoring: material PBR (Principled BSDF), graf node shader, image texture, vertex color, emisi, transparansi
Rigging: armature + bone collections (4.0+), IK chain, 20+ constraint, vertex group/weight, auto-rig, pose reset
Animation: keyframe, animasi loc/rot/scale, action, interpolasi, shape key, rigid body, gravitasi
Geometry Nodes: modifier, node, scatter instance (4.2+
interface.new_socket)UV/Printing/Batch/Analysis/IO: unwrap manual background-safe, manifold, batch rename/delete, ringkasan scene, ekspor 10 format (GLB/FBX/OBJ/USD/…)
Keamanan: validasi AST, eksekusi dengan timeout + undo, rate limiting
Multi-provider LLM: OpenAI, Anthropic, Google, DeepSeek, OpenRouter, Ollama
Verifikasi: 204 tes offline + permukaan penuh di Blender 4.2.23 nyata
Mulai Cepat
Prasyarat
Python 3.10+
Blender 4.2 LTS (atau 5.x) — addon memakai bone collections & EEVEE Next
Klien MCP (Claude Desktop, Cursor, opencode, …)
Pasang addon Blender
# Deteksi otomatis versi Blender, salin paket addon kanonik
python scripts/install.py # Linux/macOS/Windows
# atau pakai ekstensi .blender_manifest.toml langsung di Preferences > Add-onsLalu di Blender: Edit → Preferences → Add-ons → aktifkan "blender-mcp-ultra".
Addon membuka socket TCP :9876 (perintah) dan SSE MCP :9879.
Konfigurasi klien MCP
{
"mcpServers": {
"blender": {
"command": "python",
"args": ["-m", "blender_mcp.server"],
"env": {
"BLENDER_HOST": "localhost",
"BLENDER_PORT": "9876"
}
}
}
}Transport diatur BLENDER_MCP_MODE (stdio default, atau sse untuk :9879).
Uji cepat
pip install -e .
pytest tests/ -q # 204 lulus offline
python - <<'EOF'
from blender_connection import BlenderConnection
c = BlenderConnection(); c.connect()
print(c.send_command("create_object", {"type": "CUBE", "name": "Tes"}))
print(c.send_command("create_material", {"name": "Merah", "color": [1, 0, 0, 1]}))
EOFContoh Tool
Spesifikasi tool datar (bukan namespace bertitik fiktif) — sumber kebenaran di mcp_tools.py:
Domain | Tool |
Modeling |
|
Coloring |
|
Animation |
|
Rigging |
|
Geometry Nodes |
|
Scene/Render |
|
UV/Texture |
|
Batch/Analysis |
|
Printing |
|
IO |
|
Semua tool tersedia juga sebagai katalog registry list_tools/describe_tool/
call_registry_tool (dari src/tools/**) dan batch transaksional run_batch.
Chat Internal Blender (LLM)
Aktifkan di panel Axiom: pilih provider (OpenAI/Anthropic/Google/DeepSeek/
OpenRouter/Ollama), isi API key, lalu ketik perintah — mis. "buat meja kayu
dan kursi di atas lantai, warna coklat, animasikan mejanya". Asisten memakai
search_api_docs() (2.062 dokumen RST) sebelum menulis kode, memvalidasi AST,
dan menjalankan dalam undo_push.
Keamanan
Validasi AST: pola berbahaya (
os,subprocess,__import__, dunder) diblokir sebelum eksekusiSandbox: namespace terisolasi + timeout (SIGALRM) + rollback
undoRate limiting: token bucket per klien
Validasi input: pencegahan traversal path, jenis argumen diperiksa
Arsitektur
blender-mcp-ultra/
├── mcp_tools.py # Spesifikasi tunggal 138 tool (FastMCP + STDIO)
├── mcp_server.py # FastMCP (SSE :9879 / stdio)
├── blender_connection.py # Klien socket TCP :9876
├── addon/ # Addon Blender (AXIOM): _axsock, handlers, modeling,
│ # materials, rigging, animation, scene_tools, ...
├── blender_mcp/ # CLI, rst_search (dokumentasi API lokal), validasi
├── src/tools/ # ToolRegistry ~228 tool (registry bridge)
├── data/api/ # 2.062 RST dokumentasi API Blender
├── scripts/ # install.py (auto-detect versi Blender)
├── skills/ # 19+ skill
├── tests/ # 204 tes offline (stub bpy setia) + e2e
└── demo_artifacts/ # Hasil demo LLM → Blender headlessVerifikasi
Lapisan | Hasil |
Tes offline ( | 204 lulus, 105 skip (butuh binary Blender) |
Blender 4.2.23 nyata — perintah inti | 70/70 + 1 UI-only skip |
Blender 4.2.23 nyata — permukaan diperluas | 66/66 + 1 UI-only skip |
Socket E2E headless (TCP :9876) | ping, scene, create, material, animasi, workflow |
Demo LLM (gpt-5.6-sol) | 34 aksi → 3 render Cycles |
Lisensi
MIT
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Servers
- Alicense-qualityCmaintenanceTransforms Blender into an MCP server with 50+ tools for AI-driven 3D workflows, enabling complete control over objects, materials, animations, physics simulations, and rendering through natural language commands.Last updated45MIT
- Alicense-qualityDmaintenanceWindows-optimized MCP server that enables control of Blender 4.0+ through 21+ tools for scene management, object manipulation, and asset downloads from PolyHaven, Sketchfab, Hyper3D, and Hunyuan3D.Last updated13MIT
- AlicenseBqualityDmaintenanceAn MCP server that enables AI models to directly control Blender for 3D modeling, scene manipulation, and material management through natural language. It supports advanced workflows including Python code execution, viewport visualization, and integration with external asset libraries like Poly Haven and Hyper3D.Last updated221MIT
- AlicenseBqualityAmaintenanceAn MCP server that enables AI assistants to control Blender through 108 specialized tools for 3D modeling, animation, and rendering. It provides a secure, thread-safe interface to execute validated operations in Blender using natural language commands.Last updated100114AGPL 3.0
Related MCP Connectors
MCP server for AI dialogue using various LLM models via AceDataCloud
MCP server for Hailuo (MiniMax) AI video generation
MCP server for NanoBanana AI image generation and editing
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/HitamLegit6777/blender-mcp-ultra'
If you have feedback or need assistance with the MCP directory API, please join our Discord server


