# AI use cases for Cozy Clay (NomaDamas/CozyClay)

Repo: [NomaDamas/CozyClay](https://github.com/NomaDamas/CozyClay) | Use cases: 4 | Page: https://dedupedai.com/use-cases/nomadamas-cozyclay/

CozyClay is an open source browser-based previs tool for blocking 3D scenes, posing characters, and planning camera moves before sending the shots to an AI video model.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Block a scene in 3D before paying for AI video: https://dedupedai.com/use-cases/nomadamas-cozyclay/previs-scene-before-ai-video/
- Direct the studio from Claude Code or Cursor: https://dedupedai.com/use-cases/nomadamas-cozyclay/drive-studio-from-mcp-client/
- Let a coding agent drive a running studio from the terminal: https://dedupedai.com/use-cases/nomadamas-cozyclay/script-live-studio-from-terminal/
- Run and test Cozy Clay from source for contributions: https://dedupedai.com/use-cases/nomadamas-cozyclay/run-cozyclay-from-source/

## How to block a scene in 3D before paying for AI video with Cozy Clay

Repo: [NomaDamas/CozyClay](https://github.com/NomaDamas/CozyClay) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README

Lay out the set, cast, and camera in a browser tab, then export a keyframe pack for an AI video model.

**Who it is for:** A video creator planning a short shot who wants to stop paying for repeated AI video takes that miss the blocking.

**The problem:** A text prompt cannot hold where everyone stands, how they move, where the camera is, or when it cuts, so each miss costs another take.

**The outcome:** A blocked scene with timed shots and an exported first frame, last frame, greybox clip, camera, and prompt that the video model can follow.

### Steps

1. Run `npx cozyclay` (needs Node.js 22.19 or newer and a Chromium browser).
2. Open http://127.0.0.1:5180/app/ and build the set from primitives, then add rigged characters or capsule stand-ins.
3. Describe the action so the agent writes a motion spec, and check the QA results for collisions, foot slide, reach, and hand contact before you accept it.
4. Put shots on the timeline, draw dolly rails, and look through the lens to set framing.
5. Export the keyframe pack (first and last frame, greybox clip, camera, prompt) and give it to your AI video model.

### Prompt to paste into your agent

```text
Using the NomaDamas/CozyClay repo, run `npx cozyclay`, open the studio, and block a scene of <scene description> with the cast and camera shots. Export the keyframe pack for <video model>. Verify that the export includes the first and last frame, the greybox clip, the camera, and the prompt before handing it off.
```

### Caveat

Cozy Clay produces the blocking and export files, but the final video quality depends on the AI video model you send them to.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/nomadamas-cozyclay/previs-scene-before-ai-video/

## How to direct the studio from Claude Code or Cursor with Cozy Clay

Repo: [NomaDamas/CozyClay](https://github.com/NomaDamas/CozyClay) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README

Connect an MCP client to Cozy Clay so an AI agent can edit the scene while you watch the viewport.

**Who it is for:** A developer or previs artist who already works in Claude Code, Claude Desktop, or Cursor and wants the agent to build the scene for them.

**The problem:** Describing a multi-character scene and camera plan step by step in chat is slow, and you cannot see the result until you open the studio.

**The outcome:** An MCP client that edits the live studio scene, with each agent change saved as one undo step.

### Steps

1. Run `npx cozyclay` in a terminal and keep the studio open at http://127.0.0.1:5180/app/.
2. Add this to your MCP client config: `{ "mcpServers": { "cozyclay": { "command": "npx", "args": ["-y", "cozyclay", "mcp"] } } }`.
3. Restart the MCP client so it loads the Cozy Clay tools.
4. Ask the agent to block the scene, then watch the viewport as it works.
5. Undo any agent change you do not want, since each edit is one undo step.

### Prompt to paste into your agent

```text
Using the NomaDamas/CozyClay repo, connect the MCP server with `npx -y cozyclay mcp` and ask the agent to block <scene description> in the open studio. Expected output is a scene with the requested characters and camera shots visible in the viewport. Verify that the scene matches the description and that the QA results show no collisions.
```

### Caveat

The README lists Claude Desktop, Claude Code, and Cursor as examples of MCP clients, and the studio must be running for live edits to appear.

Risk, credentials: The MCP client config and any model provider key you add for the agent panel are stored on your machine, so keep those keys out of shared files.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/nomadamas-cozyclay/drive-studio-from-mcp-client/

## How to let a coding agent drive a running studio from the terminal with Cozy Clay

Repo: [NomaDamas/CozyClay](https://github.com/NomaDamas/CozyClay) | Difficulty: Advanced | Time: 30 minutes | Basis: From the README

Use cclay live commands so a coding agent can read and change a studio session with JSON in and out.

**Who it is for:** A developer who wants a scripted or coding agent to inspect and edit a scene while the studio is open in the browser.

**The problem:** Scripting scene edits by hand in the browser is repetitive, and an agent needs a way to check the live state before it acts.

**The outcome:** A coding agent that reads the status of a running studio and sends JSON commands to change the scene.

### Steps

1. Run `npx cozyclay` and open http://127.0.0.1:5180/app/.
2. Run `cclay live status --wait` to check that the studio is running and ready for commands.
3. Send the agent's JSON edits through the terminal interface described in docs/agent-cli.md.
4. Read the returned JSON to confirm the scene state after each edit.

### Prompt to paste into your agent

```text
Using the NomaDamas/CozyClay repo, run `cclay live status --wait` against the open studio and then apply <scene change> through the live terminal interface. Expected output is JSON that shows the updated scene state. Verify that the status reports a ready studio before any edit is sent.
```

### Caveat

The README gives only the status command example, so the exact edit commands must be read from docs/agent-cli.md before scripting them.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/nomadamas-cozyclay/script-live-studio-from-terminal/

## How to run and test Cozy Clay from source for contributions with Cozy Clay

Repo: [NomaDamas/CozyClay](https://github.com/NomaDamas/CozyClay) | Difficulty: Advanced | Time: 45 minutes | Basis: From the README

Clone the repo, install dependencies, start the dev server, and run the test suite.

**Who it is for:** A developer who wants to change Cozy Clay itself or check its behavior before contributing.

**The problem:** The npx install runs the packaged build, so changing the studio code needs a source checkout.

**The outcome:** A local dev server at the studio URL and a passing test run, ready for code changes.

### Steps

1. Run `git clone https://github.com/NomaDamas/CozyClay.git && cd CozyClay`.
2. Run `npm install`.
3. Run `npm run dev` and open http://127.0.0.1:5180/app/.
4. Run `npm test` to check the suite before and after your change.
5. Read AGENTS.md for the contribution rules before you open a pull request.

### Prompt to paste into your agent

```text
Using the NomaDamas/CozyClay repo, clone it, run `npm install`, `npm run dev`, and `npm test`, then report the test result. Expected output is a passing test run and the studio running at http://127.0.0.1:5180/app/. Verify that AGENTS.md has been read before any code change is proposed.
```

### Caveat

The README gives no details on the codebase structure, so contributors should read AGENTS.md and docs/studio-ui-ia.md first.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/nomadamas-cozyclay/run-cozyclay-from-source/