Skip to content
cccskills
FreePublished on GitHub

bdb-blender-mcp

Utilizes the Blender MCP server and addon to build 3D assets, apply materials, inspect scenes, and script bpy.

The English summary is being prepared. Showing the original description.

Works withBlender
See how to install

Files included (1)

  • SKILL.md6.2 KB

SKILL.md (original)

Review the instructions the agent will receive before you install.

Blender MCP — Integration and AI Agent Guide

This skill file instructs AI agents on how to interact with the Blender MCP server and addon. It details scene manipulation, Python code execution (bpy), external asset retrieval, and troubleshooting workflows.

1. Overview and Pipeline Value

The Blender MCP Server connects AI models to Blender (3.0+) via a local TCP socket bridge. In the BDB OS workflow, this integration speeds up low-poly layouts, scene prototyping, material assignment, rendering, and programmatic asset generation.

Architecture

  • Blender Addon (addon.py): Runs inside Blender, creating a socket server (default port 9876) that executes incoming requests on Blender's main thread using the Blender Python API (bpy).
  • MCP Server Bridge: A Node.js or Python-based subprocess (blender-mcp) connecting to the addon socket, presenting a clean tool layout to the AI.
  • Generative AI & API Integrations: Integrates with Poly Haven (materials, HDRIs, models), Sketchfab, Hyper3D Rodin, and Hunyuan3D for AI-assisted asset drafting.

2. System Instructions

Workflow Priorities

  1. Connect First: Ensure Blender is open and the MCP Addon is actively running (click "Connect to Claude" in the Sidebar's BlenderMCP tab).
  2. Read the Scene: Proactively query the active scene layout using introspection tools before placing meshes.
  3. Save Work Proactively: Running arbitrary script commands can crash the active Blender session. Always instruct the user to back up their project before running custom code.
  4. Clean Code Structure: When using execute_blender_code, write standard bpy script blocks. Delete reference meshes (like the default Cube) programmatically if the prompt asks for a clean slate.

3. Available Tools and API Parameters

ToolParametersDescription
get_scene_infoNoneReturns document structure, active camera, render settings, and a list of all scene objects.
create_meshtype: string, location?: float[], scale?: float[], rotation?: float[], name?: stringSpawns primitive shapes: Cube, Sphere, Cylinder, Cone, Torus, Grid, Monkey, Plane.
delete_objectname: stringRemoves a named object from the database.
modify_objectname: string, location?: float[], scale?: float[], rotation?: float[]Applies transformations to objects.
create_materialname: string, color: float[], metallic?: float, roughness?: floatCreates and configures a Principled BSDF material.
apply_materialobject_name: string, material_name: stringBinds a material to a target mesh.
execute_blender_codecode: stringEvaluates arbitrary Python scripts in Blender using the full bpy library.
search_polyhavenquery: string, type: stringQueries HDRIs, textures, or models from the Poly Haven API.
import_polyhaven_assetasset_id: string, type: stringDownloads and imports Poly Haven items directly into the viewport.
generate_hyper3d_modelprompt: stringTriggers Hyper3D Rodin API to draft a model and load it.

4. Code Recipes and Prompt Cookbook

Recipe 1: Standard Material Setup and Assignment

Create a metallic red material and assign it to a newly spawned cylinder:

// Step 1: Create the cylinder
// Tool: create_mesh(type="cylinder", location=[0, 0, 2], name="RedColumn")

// Step 2: Create red material
// Tool: create_material(name="MetallicRed", color=[0.8, 0.05, 0.05, 1.0], metallic=0.9, roughness=0.1)

// Step 3: Apply to object
// Tool: apply_material(object_name="RedColumn", material_name="MetallicRed")

Recipe 2: Run Custom Python Script via execute_blender_code

Generate a procedural spiral of spheres using bpy:

# Call tool: execute_blender_code(code="...")
import bpy
import math

# Clear default objects
bpy.ops.object.select_all(action='SELECT')
bpy.ops.object.delete(use_global=False)

# Build spiral
count = 30
radius = 5.0
height_step = 0.2
angle_step = 0.3

for i in range(count):
    angle = i * angle_step
    x = math.cos(angle) * radius
    y = math.sin(angle) * radius
    z = i * height_step
    
    # Add sphere
    bpy.ops.mesh.primitive_uv_sphere_add(radius=0.5, location=(x, y, z))
    sphere = bpy.context.active_object
    sphere.name = f"SpiralSphere_{i}"

5. Troubleshooting and Connection Details

Configuration and Ports

  • Default TCP Port: 9876 (Bypasses stdio when communicating with the Blender Addon).
  • Environment Variables:
    • BLENDER_HOST: Local/remote address of the Blender application (default: localhost).
    • BLENDER_PORT: Port number of the socket listener (default: 9876).

Telemetry Control

To disable usage telemetry, set the environment variable:

DISABLE_TELEMETRY=true

Or toggle the Consent checkbox in the Addon settings inside Blender: Edit -> Preferences -> Add-ons -> Interface: Blender MCP

Connection Verification

Ensure the Blender Addon is actively listening on port 9876:

# Check if Blender is holding the socket
lsof -i :9876

Common Errors and Fixes

  1. Cannot connect to Blender Addon at localhost:9876
    • Cause: The addon is installed but not running, or connection was closed.
    • Fix: In Blender, press N to open the sidebar, click the BlenderMCP tab, and toggle Connect to Claude ON.
  2. Missing API Keys for Hyper3D or Sketchfab
    • Fix: Add the API keys directly in the Add-on preferences page in Blender (Edit -> Preferences -> Add-ons -> Blender MCP) or inject them as environment variables (e.g. BLENDERMCP_HYPER3D_API_KEY) prior to starting the MCP client.
  3. execute_blender_code fails due to Context
    • Cause: Blender operators (bpy.ops) often require a specific active context (e.g., Edit Mode vs Object Mode, or an active selection).
    • Fix: Ensure your Python script sets the active object or overrides context if necessary before executing operators:
      bpy.context.view_layer.objects.active = target_obj
      

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

Comments are published right away. Do not include personal information such as email addresses or phone numbers. Up to 2 URLs.

0 / 1,000 characters

This form is protected by reCAPTCHA, and the Google Privacy Policy and Terms of Service apply.

More skills from this repository

Use when manage multiple local CLI agents via tmux sessions (start/stop/monitor/assign) with cron-friendly scheduling.

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

Use when a hybrid memory system that provides persistent, searchable knowledge management for AI agents (Architecture, Patterns, Decisions).

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

Reference for how BDB structures autonomous software engineering work — the seven-node dispatcher graph (Architect, TechLead, UI/UX, Engineering, Media/EventTech, Reviewer, Shipping) that /startcycle-graph actually runs. Use when you need the high-level lifecycle framing without inventing your own process.

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

Tools are how AI agents interact with the world. A well-designed tool is the difference between an agent that works and one that hallucinates, fails silently, or costs 10x more tokens than necessary. This skill covers tool design from schema to error handling.

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

Harness patterns for coding agents — memory, permissions, context engineering, delegation, skills, hooks, bootstrap.

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

agenttrail: live map of a multi-agent build in the browser: which plan component is being worked on, by which agent or harness, what is done and what is stuck. Use when a multi-agent pipeline starts (/startcycle, /startcycle-graph, /teamwork-preview) or after a plan-canvas approve, when the user asks to see what the agents are doing ("live map", "build board", "who is running now"), or by yourself whenever a multi-agent run is under way — no user prompt needed.

The English summary is being prepared. Showing the original description.

hybridlabor-api/aos6Updated Oct 16, 2026

See all skills by hybridlabor-api

Report a problem with this skill