Moondream

Provides image analysis via MCP including captioning, VQA, object detection, and visual pointing.

python

43

GitHub Stars

python

Language

5 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "colemurray-moondream-mcp": {
      "command": "uvx",
      "args": [
        "moondream-mcp"
      ],
      "env": {
        "MOONDREAM_DEVICE": "auto"
      }
    }
  }
}

Moondream MCP Server provides image analysis capabilities over the Model Context Protocol (MCP). You can run it locally to generate captions, answer questions about images, detect objects with bounding boxes, and pinpoint object coordinates, processing single images or batches efficiently. It automatically uses the best available device (CPU, CUDA, or MPS) and can read images from local files or URLs.

How to use

You connect a client to the Moondream MCP Server using one of the available MCP configurations. The server exposes tools for image captioning, visual question answering, object detection, and object pointing. You can run it via different entry points, then issue requests through the MCP client to perform operations such as captioning an image, asking questions about an image, detecting objects, or locating objects within an image. Start the server using one of the provided runtime commands, then configure your MCP client to reference that local MCP endpoint.

How to install

Prerequisites you must meet before installing the server:

Python 3.10 or higher. - PyTorch 2.0+ with device support.

Install options you can choose from to run the server locally.

# Run without installation
uvx moondream-mcp

# Or specify a specific version
uvx moondream-mcp==1.0.2

pip install moondream-mcp

git clone https://github.com/ColeMurray/moondream-mcp.git
cd moondream-mcp
pip install -e .

git clone https://github.com/ColeMurray/moondream-mcp.git
cd moondream-mcp
pip install -e ".[dev]"

Running the server and quick start

You can start the server with either the UVX runtime, the pip-installed command, or directly through Python. Choose the method that fits your setup.

# Using uvx (no installation needed)
uvx moondream-mcp

# Using pip-installed command
moondream-mcp

# Or run directly with Python
python -m moondream_mcp.server

Available tools

caption_image

Generate captions for images. You provide the image path or URL and the desired caption length, and the server returns a caption. Supports streaming and non-streaming modes.

query_image

Ask questions about the content of an image. Provide the image path or URL and your question to receive an answer.

detect_objects

Detect specific objects in an image. Specify the image path or URL and the object name to locate.

point_objects

Get coordinates of objects within an image. Provide the image path or URL and the object name to locate.

analyze_image

Multi-purpose image analysis tool. Choose an operation type such as caption, query, detect, or point and pass operation-specific parameters.

batch_analyze_images

Process multiple images in a single batch. Provide a list of image paths, the operation to perform, and any operation-specific parameters.