Ideogram MCP: generate, edit, and train from your assistant

Install

Connect any MCP client.

The server speaks streamable HTTP at https://mcp.ideogram.ai/mcp. Authentication is OAuth via your Ideogram account. The first time a client connects, a browser window opens for sign-in.

{
  "mcpServers": {
    "ideogram": {
      "transport": "http",
      "url": "https://mcp.ideogram.ai/mcp"
    }
  }
}

Add to claude_desktop_config.json. Restart the app.

claude mcp add ideogram --transport http https://mcp.ideogram.ai/mcp

Run anywhere. Sign in to your Ideogram account on first use.

Settings → Connectors → Add MCP server

URL:       https://mcp.ideogram.ai/mcp
Transport: HTTP
Auth:      OAuth (your Ideogram account)

Available in ChatGPT Plus and Pro via the desktop app's Connectors panel.

{
  "mcpServers": {
    "ideogram": {
      "url": "https://mcp.ideogram.ai/mcp"
    }
  }
}

Drop into ~/.cursor/mcp.json. Reload the editor.

{
  "mcpServers": {
    "ideogram": {
      "url": "https://mcp.ideogram.ai/mcp"
    }
  }
}

Cline reads the same JSON shape from cline_mcp_settings.json.

{
  "mcp": {
    "ideogram": {
      "type": "remote",
      "url": "https://mcp.ideogram.ai/mcp",
      "enabled": true
    }
  }
}

Add to ~/.config/opencode/config.json or a project-local opencode.json.

{
  "mcpServers": {
    "ideogram": {
      "transport": "streamable-http",
      "url": "https://mcp.ideogram.ai/mcp"
    }
  }
}

Streamable-HTTP transport. Refer to your Hermes build's config path; the JSON shape follows the standard mcpServers convention.

POST https://mcp.ideogram.ai/mcp
Authorization: Bearer <oauth-token>
Content-Type: application/json

{
  "jsonrpc": "2.0",
  "method": "tools/list",
  "id": 1
}

Any client that speaks MCP streamable-HTTP works. Discover OAuth via /.well-known/oauth-authorization-server.

Workflows

See what teams build with it.

Each of these is a single conversation. The agent chains the right tools without anyone writing glue code.

Asset packs from a brief

Hand the agent a one-paragraph campaign brief. It returns a hero, four social cards, and two print posters in a single conversation, saved into a named collection.

generate_images_bulk
create_collection
upscale_image

Catalog cleanup

Feed product photos into the chat. The agent removes backgrounds, reframes for marketplace ratios, and uploads the cleaned assets back to a shared collection.

edit_image
reframe_image
upload_image

Train and iterate

Drop a folder of brand references. The agent creates a dataset, uploads the assets, kicks off training, and pings you when the model is ready to use.

create_dataset
upload_dataset_assets
train_model

Questions

Frequently asked

Which clients are supported? +

Anything that speaks the MCP streamable-HTTP transport. This includes Claude (desktop and code), ChatGPT, Cursor, Cline, OpenCode, and any custom client that implements the protocol.

How does authentication work? +

Your client opens a browser to sign you into your Ideogram account via OAuth. The MCP server acts on your behalf using the resulting session, with no API key handling required.

Does the Ideogram MCP server use my plan's credits? +

Yes. Usage through the MCP server draws from the same subscription as the Ideogram web app. There is no separate billing for MCP requests.

When should I use the Ideogram REST API instead of the MCP server? +

MCP is built for agents that act on a user's behalf with OAuth. For server-to-server pipelines under your own control, the REST API at developer.ideogram.ai is a better fit.

Add Ideogram to any AI agent