Blender · Model Context Protocol
Blender, driven by your AI assistant
Connect Blender to Claude, Cursor or any MCP client. Ask what is in the scene, change it with typed operations rather than generated scripts, and export control passes that let a blocked-out shot dictate a video model’s camera and timing.
Install the add-on, then sign in to your Draftless account from inside Blender.
Your assistant, in your scene
Ask what is in the shot and get a real answer — objects, lights, camera, timeline, selection — read live out of the Blender you have open, not a guess from a screenshot.
Typed operations, not generated scripts
Every change arrives as a validated operation with a known shape. An assistant cannot invent a bpy call that deletes your work, because free-form Python never reaches your session.
Blocked-out shots become video
Render the control passes from a rough layout and hand them to a video model. The 3D motion dictates the camera and the timing instead of hoping a prompt lands them.
Sign in once, from inside Blender
The add-on signs in with a code you confirm in your browser — no key to paste. That one sign-in covers both the live connection to your assistant and the upload of exported passes.
How it fits together
- 1. The add-on runs inside Blender. It opens a connection out to the hosted bridge. Your scene never leaves your machine unless you export it.
- 2. The bridge relays your assistant. It forwards typed tool calls to your session and nothing else. It never runs Blender Python itself.
- 3. Exports land in your library. Control passes upload straight into Draftless, ready to drive a video model alongside your brand kit and characters.
Domande, con risposta
What is the Draftless Blender MCP integration?
It lets an AI assistant drive the Blender you already have open. The Draftless add-on runs inside Blender and connects to a hosted bridge; any MCP client — Claude, Cursor, ChatGPT — can then inspect the scene, apply changes and render preview passes by asking for them in plain language.
Does it run arbitrary Python in my Blender?
No. The bridge accepts typed scene operations only — add an object, set a transform, set a light, move the camera, set the timeline, insert a keyframe, apply a motion preset. Free-form Python is rejected rather than executed, so generated code never runs in your session.
Which Blender version does it need?
Blender 4.2 LTS or newer. The current add-on (0.2.1) is tested on Blender 4.2 LTS and Blender 5.2.
What does it export for video generation?
A control-pass bundle from your blocked-out shot: beauty, depth, normals, optical flow and object masks as video, plus the first frame and a camera sidecar. Depth and the first frame are what the current video adapter needs; the rest are exported once so new adapters need no add-on update.
Is it available yet?
Yes. Download the add-on from the Blender plugin page, install it in Blender, and sign in to your Draftless account from inside Blender. Then connect it to the hosted bridge and add the bridge to your MCP client.