Agnes AI MCP Server

An MCP server that provides text-to-image and text-to-video generation via the Agnes AI API.

Features

• Text-to-Image: Generate images from text prompts (models: agnes-image-2.0-flash, agnes-image-2.1-flash)

• Text-to-Video: Generate videos from text prompts or images (model: agnes-video-v2.0)

• Video Status Check: Poll async video generation tasks

• Auto-download: Automatically downloads generated files locally

Related MCP server: TensorsLab MCP Server

Installation

# Via uvx (recommended)
uvx agnes-mcp

# Via pip
pip install agnes-mcp

Configuration

Add to your MCP client config (e.g. Claude Desktop, Codex, etc.):

{
  "mcpServers": {
    "agnes-mcp": {
      "command": "uvx",
      "args": ["agnes-mcp"],
      "env": {
        "AGNES_API_KEY": "your-api-key-here"
      }
    }
  }
}

Or for TOML-based configs (e.g. Codex):

[mcp_servers.agnes-mcp]
command = "uvx"
args = ["agnes-mcp"]

[mcp_servers.agnes-mcp.env]
AGNES_API_KEY = "your-api-key-here"

Environment Variables

Tools

text_to_image

Generate an image from text. Supports agnes-image-2.0-flash and agnes-image-2.1-flash.

text_to_video

Generate a video from text (and optional image). Supports agnes-video-v2.0. Async operation - polls until completion.

check_video_status

Check the status of a video generation task by video_id or task_id.

Get an API Key

1. Visit Agnes AI

2. Create an account

3. Generate an API key from the console

License

MIT