Blender AI MCP is a Model Context Protocol (MCP) server developed by Patrykiti, designed to integrate AI capabilities directly into Blender for 3D creation. This free tool connects to a running Blender instance, where its addon handles operations on Blender's main thread. The MCP server component manages routing, validation, discovery, and structured responses, enabling AI clients like Claude, ChatGPT, and Codex to interact with Blender. It supports various workflows, from goal-first guided sessions to iterative modeling and basic object manipulation, catering to those interested in creative automation and agent-based 3D design.

What does Blender AI MCP connect to?

Blender AI MCP functions as an MCP server for Blender, establishing a connection to a live Blender instance. Its Blender addon component executes operations directly on Blender's main thread. Concurrently, the MCP server component performs essential tasks such as routing requests, validating inputs, facilitating discovery of tools, and generating structured responses. This architecture allows AI clients to interact with and control Blender, making it a central hub for AI-driven 3D modeling and automation within the Blender environment.

Core features and tools

The program exposes specific callable tools including router_set_goal(...) and router_get_status(...). It also offers reference_compare_stage_checkpoint and reference_iterate_stage_checkpoint for iterative modeling. Future modeling tools planned include create_primitive, transform_object, and add_modifier. These tools enable actions such as creating basic shapes (Cube, Sphere, Cylinder, Plane), moving, rotating, and scaling objects, and applying modifiers like Bevel, enhancing automated 3D design capabilities through an AI interface.

Who is it for?

Blender AI MCP is for individuals keen on AI-assisted 3D creation, creative automation, and agent-based workflows. It also serves those interested in understanding practical MCP server architecture. Its capabilities support goal-first guided sessions and iterative, reference-guided 3D modeling. This is not the right fit if you require an explicit authentication model for server access.

How to get started

No credentials are required. The Blender addon starts a local RPC server on port 8765, and the MCP server component, typically run via a Docker image, is configured using environment variables such as BLENDER_AI_TMP_INTERNAL_DIR, ROUTER_ENABLED, and BLENDER_RPC_HOST to manage its operation and connection settings.

Use cases

• Goal-first guided sessions where the system understands the model's objective before executing actions.
• Iterative, reference-guided 3D modeling using tools like reference_compare_stage_checkpoint and reference_iterate_stage_checkpoint for deterministic stage comparison and correction.
• Basic 3D object manipulation, including creating primitives (Cube, Sphere, Cylinder, Plane), transforming objects (move, rotate, scale), and adding modifiers (e.g., Bevel).

Blender AI MCP offers a valuable bridge between AI agents and 3D creation in Blender, enabling automated and guided modeling workflows. Its design supports various AI clients and provides foundational tools for object manipulation and scene management. The open approach to integration makes it accessible for experimentation in creative automation.