baiblinocun1986/blender-ai-copilot-hub

Overview

blender-ai-copilot-hub is a Blender extension that puts OpenAI, Anthropic Claude, Google Gemini, OpenRouter, OpenCode Zen, Ollama and custom OpenAI- or Anthropic-compatible endpoints behind one sidebar panel. Connections are saved as profiles, so separate work and personal keys can be switched without retyping them.

The copilot reads the scene through 27 supervised tools and turns a request into ordered operations that require approval. It also exposes the same toolset to external MCP clients through a local authenticated endpoint. Typical users are 3D artists and technical designers who want to drive Blender with an LLM while keeping step-by-step control.

Quick Start

1. Download mcp-for-blender-pro.zip from the latest release.
2. In Blender, open Edit → Preferences → Get Extensions → menu → Install from Disk and select the ZIP.
3. Enable mcp-for-blender-pro, then open 3D Viewport → N → MCP Pro.
4. Choose a provider, add connection details and select a model that supports tool calling.
5. Send a message or click Open Studio Window.

The README suggests prompts such as inspecting the scene and building a product display with a pedestal, a camera and three lights. A viewport snapshot should be returned after the edit is approved.

Configuration

Custom endpoints expect a base URL such as https://your-service.example/v1, not a full completion path, plus a protocol choice: OpenAI Chat Completions, Anthropic Messages or Ollama. When the live catalog does not load, the model ID can be entered manually. The MCP server is off until enabled in preferences; once on, clients use http://127.0.0.1:9876/mcp with a bearer token. Workflow mode can be a single assistant or a team chain with manager, workers and reviewers.

Constraints

The README lists Windows x64 and Blender 4.2+ as requirements, and the interface supports English, German, French, Spanish and Chinese. Keys are stored in Blender user settings rather than the scene file. Models must support tool calling to edit the scene; otherwise the assistant can chat but not act. The README does not describe macOS or Linux support, per-operation approval granularity, or how scene snapshots are sent to each provider.

HotRepo's take

HotRepo recorded 100 stars and 1 fork on the new-project list on 2026-11-13, with 100 stars gained in roughly 17 hours from the 2026-11-12 19:32 UTC baseline. That single snapshot shows a fast start but does not establish sustained adoption. The repository has no detected programming-language code while the product is distributed as a precompiled Blender ZIP, so readers should check the release contents and the relationship between the binary and the stated MIT source before relying on it.