# Connect your coding agent to OTOY Studio

Connect once, sign in with your OTOY Studio account, and your agent can generate images, video, 3D models, audio and Gaussian splats and save them into your project. There are no API keys, and runs use the same credits as the app.

**MCP endpoint:** `https://canvas-mcp.otoy.studio/mcp` (streamable HTTP, OAuth sign-in)

You need an OTOY Studio account. If you don't have one, [start free](https://app.otoy.studio): no credit card, 100 credits a month.

## If you are a coding agent

Add the server using the steps for your client below. You cannot finish sign-in yourself, so tell the user when to sign in, and that they may need to restart you first. Before a run, use `get_pricing` and `get_balance` and tell the user the price. Never call `share_media` without asking, because share links are permanent.

## Connect your agent

### Claude Code

1. Add the server from your terminal.

   ```bash
   claude mcp add otoy-studio --scope user --transport http https://canvas-mcp.otoy.studio/mcp
   ```
2. If Claude Code is already running, restart it so it picks up the new server.
3. Run `/mcp`, select `otoy-studio`, choose Authenticate, and sign in with your OTOY Studio account in the browser window that opens.

### Codex

1. Add the server from your terminal. Codex opens the sign-in in your browser straight away.

   ```bash
   codex mcp add otoy-studio --url https://canvas-mcp.otoy.studio/mcp
   ```
2. Sign in with your OTOY Studio account. If you need to sign in again later, run `codex mcp login otoy-studio`.

### Cursor

1. Add the server to `~/.cursor/mcp.json`, or to `.cursor/mcp.json` for a single project.

   ```json
   {
     "mcpServers": {
       "otoy-studio": { "url": "https://canvas-mcp.otoy.studio/mcp" }
     }
   }
   ```
2. Sign in with your OTOY Studio account when Cursor prompts you.

### VS Code

1. Open the Command Palette, run **MCP: Add Server** and choose **HTTP**.
2. Paste the endpoint URL.

   ```bash
   https://canvas-mcp.otoy.studio/mcp
   ```
3. Sign in with your OTOY Studio account when VS Code prompts you.

### Gemini CLI

1. Add the server from your terminal.

   ```bash
   gemini mcp add --scope user --transport http otoy-studio https://canvas-mcp.otoy.studio/mcp
   ```
2. Start Gemini CLI, run `/mcp auth otoy-studio`, and sign in with your OTOY Studio account.

## Prompts to try

- **Hero video:** Look at this landing page and generate a short, looping hero video that fits it. Tell me the price before you run anything, then save the clip to the project and wire it in.
- **Placeholder art:** Find the placeholder images in this project and replace them with generated ones that match the style of the site. Check my credit balance first and show me the total cost.
- **Game assets:** Generate a low-poly treasure chest as a 3D model, compare the price of two 3D models first, and save the file to assets/models.
- **Voiceover:** Write a 20-second voiceover script from this README, generate it with ElevenLabs, and save the audio to the project.

## What your agent can do

- **Find a model and check the price:** Search the live catalogue, read each model's inputs, and check the price and your balance before a run. Tools: `list_endpoints`, `get_endpoint_docs`, `get_endpoint_schema`, `get_pricing`, `get_balance`.
- **Run jobs in the background:** Fast jobs return right away. Long video and 3D renders keep running while your agent works on something else. Tools: `run_endpoint`, `check_job`, `list_jobs`, `cancel_job`, `forget_job`, `render_job_media`.
- **Bring in your own files:** Upload local files as inputs, such as a product photo or a reference frame, and reuse them later. Tools: `upload_file`, `request_upload_url`, `confirm_upload`, `list_files`, `get_file_url`.
- **Look back at past runs:** Search past generations by prompt or model and see what each one cost. Share an output when you want to. Tools: `list_generations`, `share_media`.

## Common questions

### Do I need an API key?

No. You sign in with your OTOY Studio account through OAuth, so there is no key to create, paste or rotate.

### What does it cost?

Runs use the same credit balance as the app, and your agent can check a model's price and your balance before it runs anything. Failed and cancelled runs are refunded. The free plan includes 100 credits a month with no credit card. See plans and prices at https://otoy.studio/pricing.

### Which models can my agent use?

The same catalogue as the app: more than 100 endpoints across image, video, 3D, audio and upscaling, including GPT Image 2, Nano Banana Pro, Veo 3.1, Kling, Hyper3D Rodin and ElevenLabs. Your agent picks one, or uses the one you name. See every model and price at https://otoy.studio/models.

### Can I run it on a remote machine or in CI?

Sign-in opens a browser and returns to your local machine, so connect the server somewhere you can open one.

### Are outputs private?

Outputs are saved to your OTOY Studio files. Sharing is opt-in, but a share link is permanent and cannot be revoked, so ask your agent to share only what you want public.

### Does it work with other clients?

Any MCP client that supports streamable HTTP and OAuth can use the same URL: https://canvas-mcp.otoy.studio/mcp

## Related

- [All AI models and prices](https://otoy.studio/models)
- [Pricing](https://otoy.studio/pricing)