Seedream

You can run Seedream MCP to generate AI images through a local or remote MCP client. This server connects to Seedream’s APIs to produce text-to-image, image-to-image, and related image generation workflows, enabling you to automate creative tasks with a simple command-line or Docker setup.

How to use

You will run the MCP server using a client that supports MCP connections, such as uvx. Start by providing your API key to authenticate with Seedream’s MCP. You can then invoke image generation workflows like text-to-image, image-to-image, multi-image fusion, and sequential image generation through the client’s MCP endpoints. The following examples assume you have a running MCP server configured for Seedream.

How to install

Prerequisites: you need a working environment with Node.js or a runtime that can run the MCP client, and you should have access to the internet to fetch the Seedream MCP tooling.

Option 1: uvx one-click startup (recommended) — just run the MCP client directly from the repository.

# Directly launch from the GitHub repository
uvx git+https://github.com/tengmmvp/Seedream_MCP --api-key your_api_key_here

# Or clone first, then start
git clone https://github.com/tengmmvp/Seedream_MCP
cd Seedream_MCP
uvx . --api-key your_api_key_here

Option 2: Docker Compose — run Seedream MCP in a containerized environment.

# Download docker-compose file
curl -O https://raw.githubusercontent.com/tengmmvp/Seedream_MCP/main/docker-compose.yml

# Start the service with your API key set in the environment
ARK_API_KEY=your_api_key_here docker-compose up -d

Option 3: Claude Desktop configuration — configure the MCP server by adding a small JSON snippet to Claude Desktop so you can use Seedream’s MCP features directly from Claude.

{
  "mcpServers": {
    "seedream": {
      "command": "uvx",
      "args": [
        "git+https://github.com/tengmmvp/Seedream_MCP",
        "--api-key",
        "your_api_key_here"
      ]
    }
  }
}

Additional setup notes

You can customize the default model, image size, and other options when you start Seedream MCP. The server supports different transmission modes when connecting via an MCP client, including stdio, SSE, or streamable-http depending on your setup.

Configuration and troubleshooting

If you need to troubleshoot, ensure your API key is valid and that your MCP client is configured to reach Seedream MCP. For Docker-based runs, verify that ARK_API_KEY is set in your environment before starting the containers.

Notes and recommended practices

Keep your API keys secure and do not expose them in public scripts. For automated generation tasks, consider using a configuration file to store default size, watermark preferences, and output paths to ensure consistent results across runs.

Available tools

seedream_text_to_image

Generate an image from a textual prompt with configurable size, watermark, and output options.

seedream_image_to_image

Create a new image by transforming an input image guided by a text prompt.

seedream_multi_image_fusion

Fuse multiple input images into a single generated image guided by a prompt.

seedream_sequential_generation

Generate a sequence of images, enabling storyboard-style or group image outputs.

seedream_browse_images

Browse local image assets to prepare inputs for generation.