Claude / Codex ↔ Blender Integration (MCP-based)
MCP-based integration, where LLM models (Claude, Codex, OpenAI) control Blender through Blender's Python API. The project enables agent-based workflows that combine natural language control, context management, code generation, Blender runs, and result validation. Designed to explore how AI can be integrated into creative and technical 3D work in a controlled, safe, and repeatable manner.
Focus areas:
- Agent workflows & tool orchestration
- Blender Python API automation
- Guardrails & rollback safety
- Prompt → 3D / GLB pipeline
Tech Stack
Mac Desktop app logo
Mac Desktop app setup screen
Beta testing the connection
Utilized the integration for this model
Procedural / Technical Test Prompt: Create a procedural medieval watchtower generator in Blender. Requirements: - tower height parameter - random seed parameter - optional ruined mode - procedural stone material - staircase generation - modular windows and arches - defensive wooden platform - surrounding terrain generation Build this as reusable Geometry Nodes where possible. After generation: - create 5 unique tower variants - arrange them naturally in a scene - optimize geometry - generate UVs - create LOD versions - export all towers as separate GLB assets. Claude Desktop Opus 4.7.