Create a product image
Use Cuta to create a studio product photo of an amber perfume bottle on warm sand, with soft side lighting. Find a suitable image model and quote a square image before generating.
Cuta MCP
Your ideas. Your AI assistant.
A whole creative studio.
Bring Cuta into the conversation. Create images, videos and audio with your favorite AI assistant, using the same projects, assets and credits.
https://api.cuta.app/mcpSign in with your Cuta account. No API key needed.
Images
Video
Audio
CutaYour creative tools, togetherCreative examples from Cuta. Your results depend on the model and prompt.
One connection
Make yourself at home in any assistant
Choose your client, add Cuta, and sign in to start creating.
Connect Cuta to Claude
Use a custom connector in Claude on the web or desktop.
Add a custom connector
Open Settings → Connectors in Claude and choose to add a custom connector. Name it Cuta.
Paste the server URL
Use the Cuta MCP URL below, then add and connect the server.
Sign in and create
Sign in to your Cuta account and review the requested permissions. Enable Cuta in your conversation and describe what you want to make.
MCP server URL
https://api.cuta.app/mcpPrefer the terminal? Connect with Codex CLI or Claude Code
With your chosen client already installed, run its command in your local terminal. These commands connect the client to the same Cuta MCP server.
Codex CLI
codex mcp add cuta --url "https://api.cuta.app/mcp"
codex mcp login cutaClaude Code
claude mcp add --transport http cuta "https://api.cuta.app/mcp"In Claude Code, run /mcp after adding the server to complete OAuth sign-in. In Codex, use the login command above. Review your Cuta permissions in the browser, then return to your assistant.
Your studio, within reach
From the first idea to the next iteration
Keep the conversation going while Cuta handles the creative work. Your assistant can find a model, quote a generation and retrieve the finished result.
Describe the image. Keep refining it.
Ask your assistant to find a suitable image model and turn your brief into a generation. Bring existing Cuta assets into the same project as references for your next version.
Your assistant checks the model's supported settings and credit quote before submitting a task.
Explore image creationWorkflow exampleCuta MCP
Use Cuta to create a studio product photo of an amber perfume bottle on warm sand, with soft side lighting. Find a suitable image model and quote a square image before generating.
Start a conversation
A few words. A new starting point.
Copy a prompt into your connected assistant. It will help you choose a project, check available models and confirm the generation cost.
Turn an image into a video
Find the latest product image in my Cuta project and make a short cinematic video with a slow camera push-in. Show me a compatible model and the credit quote first.
Set the mood with music
Use Cuta to find a music model for a calm, instrumental soundtrack for my product film. Show me the supported duration and credit cost before creating it.
Find your next starting point
List my Cuta projects so I can choose one, then show me the recent generation tasks in that project.
Build on an earlier idea
Show the recent images in my Cuta project. Help me choose a reference and a compatible model for a new version with the same visual style.
Check your creative budget
How many Cuta credits do I have, and when do they expire? Help me get a quote for my next generation without submitting a task yet.
Made for your workflow
One connection. Ten creative tools.
Your assistant discovers these tools from Cuta. Available models and their settings come from the live model catalog, so you can choose what fits each project.
Explore the model catalog- cuta_get_canvas
- Read a canvas and continue from its existing nodes
- cuta_search_models
- Find models by category and capability
- cuta_get_model
- Read model settings and supported inputs
- cuta_list_projects
- Browse your existing projects
- cuta_get_credit_balance
- Check credits and upcoming expiry
- cuta_list_assets
- Find reference assets in a project
- cuta_quote_generation
- Get a quote before generating
- cuta_generate
- Submit an image, video or audio task
- cuta_get_task
- Follow progress and retrieve results
- cuta_list_tasks
- Browse project generation history
A little more about Cuta MCP
What is Cuta MCP?
MCP is an open protocol that lets AI assistants use external tools. Cuta MCP connects your assistant to Cuta's models, generation tasks, projects, assets and credits.
Which assistants can I connect?
Use the guides above for Claude, ChatGPT, Cursor and Codex. Other clients need Streamable HTTP, OAuth with PKCE and dynamic client registration. Custom connection availability may depend on your client's plan or workspace policy.
Do I need an API key?
No. Add the MCP server URL and sign in with your Cuta account through OAuth. Review the requested access on Cuta before approving the connection.
How are generations charged?
Generations use the same Cuta task and credit system as the website, including applicable plan discounts. The assistant can request a quote before submitting. Connecting your account and browsing your projects do not consume generation credits.
Where do my results go?
Completed results are saved to the Cuta project selected for the task. Your assistant retrieves media links when the task finishes. Whether images or videos display inline depends on the client.
Can I use my own reference files?
Yes. Upload your files to the relevant Cuta project first, then ask your assistant to find and use them. Reference types depend on the selected model. External URLs and attachments sent only to your assistant are not imported automatically.
Which models are available?
Ask your assistant to search Cuta's live catalog. It can read each model's supported parameters and get a quote. Availability depends on the models enabled on Cuta; the connection does not guarantee every model supports every input or feature.
What access am I granting?
Read access covers your projects, assets, task history, credit balance and model information. Generation access lets the client create tasks that spend your Cuta credits. Review the requesting client's name and permissions on the authorization page. You can disconnect Cuta in your client's connection settings.
Can I use Cuta from a terminal?
Yes. Connect the installed Codex CLI or Claude Code client using the commands above. They use the same MCP server and Cuta account. There is no separate Cuta CLI installation required.
What if the connection does not finish?
Check the server URL and confirm your client supports remote MCP and OAuth. If sign-in expires, start a new connection from the client. A managed workspace may need administrator approval. For an unsupported hosted callback, contact Cuta support with the client name and callback URL, without sending passwords or tokens.
Keep your next idea in the conversation.
Connect your Cuta account and let your assistant help you take it from here.
Connect Cuta