Skip to main content

Documentation Index

Fetch the complete documentation index at: https://www.recraft.ai/docs/llms.txt

Use this file to discover all available pages before exploring further.

Recraft supports the Model Context Protocol (MCP), enabling AI agents such as Claude, Cursor, and other MCP-compatible clients to interact with Recraft’s image generation and editing tools through a standardized interface. By connecting Recraft’s MCP server to your client, you can generate and edit high-quality raster and vector images directly from your agent environment without switching tools. Recraft offers two ways to connect:
  • Remote MCP server — a hosted server at https://mcp.recraft.ai/mcp that requires no local installation. Authenticate via OAuth with your Recraft account and connect directly from any MCP-compatible client. Usage is billed against your Recraft subscription credits (the same balance used in the Recraft Studio app). See Remote MCP server for setup instructions.
  • Local MCP server — a self-hosted server you run on your machine. Useful if you want to store generated images locally or need full control over the setup. Usage is billed against your API units balance (the same balance used by the Recraft API) and requires a Recraft API key. See Setup for installation options.
The remote and local MCP servers draw from different balances. The remote server consumes subscription credits from your Recraft plan, while the local server consumes API units purchased separately. Choose the option that matches the balance you want to use.

Supported operations

Both Recraft’s MCP servers expose the following operations:
  • Raster and vector image generation
  • Raster and vector image transforms
  • Creating and using custom styles
  • Vectorization of raster images
  • Background removal and replacement
  • Upscaling of raster images
For the source code of the Local MCP Server, visit Recraft MCP GitHub.

Choosing remote vs. local

Use this comparison to decide which option fits your workflow:
Remote MCP serverLocal MCP server
InstallationNone — hosted at https://mcp.recraft.ai/mcpRequires Node.js, or install via DXT / Smithery
AuthenticationOAuth 2.0 with your Recraft accountRecraft API key
BillingSubscription credits (same balance as the Recraft Studio web app)API units (same balance as the Recraft API)
Result storageHosted by Recraft; clients receive URLsLocal directory (default) or hosted; configurable
Best forFast setup, Claude App / web clientsLocal file output, custom environments, users on a pay-as-you-go API plan
See Remote MCP server or Local MCP server for setup details.

Supported clients

Any MCP-compatible client that supports the Streamable HTTP transport and OAuth can use the remote server; any client that supports stdio-based servers can use the local server.

Troubleshooting

Remote MCP server

  • OAuth window doesn’t open or loops back. Clear the cached OAuth state in your client and reconnect. For mcp-remote, delete ~/.mcp-auth/ (or the equivalent directory for your OS) and restart the client.
  • Tools don’t appear in the client. Confirm the connector is enabled in your client’s tools menu. In Claude Code, run /mcp to verify the server is connected and authenticated.
  • “Insufficient credits” errors. Top up your subscription credits on the Recraft web platform. Remote MCP calls deduct from this balance, not from your API units.

Local MCP server

  • 401 Unauthorized from the API. Check that RECRAFT_API_KEY is set in the client config and that your API units balance is greater than zero on the API page.
  • Server doesn’t start. Verify Node.js is installed and that the config file path is correct for your client. Restart the client after editing the config so it re-reads the MCP server list.
  • “API units depleted” errors. Buy additional API units from your profile page. API units are separate from subscription credits.

Either server

  • Generation succeeds but no image preview is shown. Some clients impose a size limit on inline images. Retry with a smaller size, or open the returned URL directly.
  • Tools call appears to hang. Image operations can take up to a few minutes for large outputs or upscales. Wait for the call to time out before retrying.