Imaginode MCP server
{
"mcpServers": {
"imaginode": {
"type": "http",
"url": "https://imaginode.ai/api/mcp"
}
}
}{
"mcpServers": {
"imaginode": {
"type": "http",
"url": "https://imaginode.ai/api/mcp"
}
}
}{
"mcpServers": {
"imaginode": {
"type": "http",
"url": "https://imaginode.ai/api/mcp"
}
}
}{
"servers": {
"imaginode": {
"type": "http",
"url": "https://imaginode.ai/api/mcp"
}
}
}{
"mcpServers": {
"imaginode": {
"type": "http",
"url": "https://imaginode.ai/api/mcp"
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsonversion 1.0.2
What it does
Imaginode MCP Server connects Claude, Cursor, or any MCP client to the imaginode.ai image and video generation API. It exposes access to 48+ AI models (image and video) through a single credit balance, with automatic refunds on failed generations. Intended for developers who want to generate media from within AI assistant workflows.
- Generate an image from a text prompt using one of 48+ models (Flux, GPT Image, Nano Banana, etc.) and receive a direct URL.
- Start a video generation job, receive a request ID, and poll for completion status until the output URL is available.
- Browse available image and video models with their credit prices before choosing which to use for a generation.
Tools this server exposes
| Tool | Does |
|---|---|
| list_models | Returns the catalogue of available image and video models with their credit prices and billable options. |
| generate_image | Generates an image from a prompt and returns the result URL directly; defaults to Nano Banana 2. |
| generate_video | Starts a video generation job and returns a request_id and the exact credit cost; defaults to Seedance 2.0. |
| get_generation_status | Checks the status of a generation by request_id, returning running, completed with output URLs, or failed with automatic refund. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.