Ideogram Agents (MCP)
Product · Developer tool by Ideogram
Bring visual creation into your agent workflow with Ideogram MCP.
Ideogram Agents (MCP) connects Ideogram's image generation and editing tools to AI assistants like Claude and ChatGPT. It enables developers to automate creative workflows, including bulk generation and model training, directly within their preferred agent environment.
- Languages
- English
Category: AI Agents
- Platforms
- Web, macOS, Windows, Linux
- Integrations
- Claude, ChatGPT, Cursor, Cline, OpenCode, Hermes
- Deployment
- Remote server at https://mcp.ideogram.ai/mcp
- Protocol
- Model Context Protocol (MCP)
- Transport Type
- Streamable HTTP
- Server URL
- https://mcp.ideogram.ai/mcp
- Connects Ideogram to Claude, ChatGPT, and Cursor
- Automates bulk generation and catalog cleanup
- Enables model training from within agent chats
- Uses OAuth authentication with shared plan credits
Overview
Ideogram Agents (MCP) is a developer tool that connects Ideogram's generative media platform to AI assistants and agents using the Model Context Protocol. It allows businesses and creators to generate, edit, and train custom models directly within tools like Claude, ChatGPT, and Cursor without switching applications. The server speaks streamable HTTP and uses OAuth for secure authentication, acting on behalf of the user's existing Ideogram account. Agents can automate complex production tasks such as creating bulk asset packs from a single brief or performing catalog cleanup by removing backgrounds and reframing images. This integration ensures that visual creation remains part of the agent's workflow while drawing from the user's shared pool of subscription credits.
Creative Agent Workflows
- Agent Workflow Integration: The server connects Ideogram to Claude, ChatGPT, Cursor, and other MCP-compatible clients so agents can generate and edit images without breaking context.
- Bulk Asset Generation: Agents can generate a hero image, four social cards, and two print posters from a single one-paragraph campaign brief.
- Automated Catalog Cleanup: The agent can automate catalog cleanup by removing backgrounds and reframing images for marketplace ratios before uploading them to a collection.
- Model Training Automation: Users can drop a folder of brand references for an agent to create datasets, upload assets, and kick off custom model training.
Developer Implementation
- Streamable HTTP Transport: The server speaks streamable HTTP at a dedicated endpoint for any client that implements the protocol.
- OAuth Authentication: Authentication is handled via OAuth through the user's Ideogram account, opening a browser window for sign-in on the first connection.
- Shared Credit Usage: Usage through the MCP server draws from the user's existing Ideogram subscription credits with no separate billing for requests.
- Tool Chaining: The server allows agents to chain multiple tools, such as image generation, upscaling, and collection creation, in a single conversation.
Use cases
- Creative teams generate entire social media asset packs from a single campaign brief within their AI assistant.
- Ecommerce managers automate background removal and reframing for thousands of product photos using an agent.
- Developers train custom branded models by dropping a folder of references into a chat with an MCP-compatible agent.
- Designers iterate on existing visuals by describing edits to an agent that uses Ideogram's precise editing tools.
Specifications
- Platforms
- Web, macOS, Windows, Linux
- Editors and tools
- Cursor, Claude Desktop, ChatGPT Desktop, Cline, OpenCode, Hermes Bot
- Integrations
- Claude, ChatGPT, Cursor, Cline, OpenCode, Hermes
- Deployment
- Remote server at https://mcp.ideogram.ai/mcp
- Protocol
- Model Context Protocol (MCP)
- Transport Type
- Streamable HTTP
- Server URL
- https://mcp.ideogram.ai/mcp
- Authentication
- OAuth via Ideogram account
- Supported Clients
- Claude, ChatGPT, Cursor, Cline, OpenCode, Hermes
- Billing Integration
- Shared with Ideogram subscription credits
How it works
- Configure the Client: Add the Ideogram MCP server URL and transport type to your client's configuration file, such as claude_desktop_config.json or mcp.json for Cursor.
- Authenticate via OAuth: The first time the client connects, a browser window opens for you to sign into your Ideogram account via OAuth.
- Interact with the Agent: Provide a brief or instructions to your assistant; the agent will pick the right tools from the catalog to generate, edit, or train.
FAQ
Which clients are supported?
Ideogram MCP supports any client that speaks the MCP streamable-HTTP transport, including Claude (desktop and code), ChatGPT, Cursor, Cline, and OpenCode.
How does authentication work?
Authentication is handled via OAuth; your client opens a browser for you to sign into your Ideogram account, and the server acts on your behalf using that session.
When should I use the REST API instead of the MCP server?
MCP is designed for agents acting on a user's behalf via OAuth, while the REST API is better suited for server-to-server pipelines under your own control.
Added by AntSpace from public sources (ideogram.ai).