vision-mcp

MCP server that provides vision analysis tools via an OpenAI-compatible API. Enables non-vision coding models to "see" images by routing image analysis to a separate vision-capable model.

Setup

Prerequisites

• An OpenAI-compatible vision API endpoint (e.g., OpenAI, opencode.ai, or any provider supporting /v1/chat/completions with image_url content blocks)

Installation & Usage with uvx

The recommended way to run the server is via uvx — no manual installation needed:

uvx vision-mcp-server

Or install globally with uv:

uv tool install vision-mcp-server
vision-mcp-server

Environment Variables

The server communicates over stdio using the MCP JSON-RPC protocol.

Tools

vision_analyze_image

Analyzes a single image file with an optional prompt.

| Parameter | Required | Description | |---|---|---|---| | image_path | Yes | Path to a local image file (PNG, JPEG, GIF, WebP) | | prompt | No | Guiding prompt (default: "Describe this image in detail.") |

vision_compare_images

Compares 2 to 8 image files simultaneously with a required prompt.

| Parameter | Required | Description | |---|---|---|---| | image_paths | Yes | List of 2-8 local image file paths | | prompt | Yes | Prompt describing what to compare |

Coding Agent Integration

opencode / Claude Code

Add to your project's .mcp.json:

{
  "mcpServers": {
    "vision": {
      "command": "uvx",
      "args": ["vision-mcp-server"],
      "env": {
        "VISION_API_BASE": "https://api.openai.com/v1",
        "VISION_API_KEY": "${VISION_API_KEY}",
        "VISION_MODEL": "gpt-4o"
      }
    }
  }
}

Note: The server installed via uvx/uv tool install creates the entry point vision-mcp-server (derived from the project name vision-mcp and script name server).

Supported Image Formats

• PNG (.png)
• JPEG (.jpg, .jpeg)
• GIF (.gif)
• WebP (.webp)

Maximum file size: 20 MB.

License

MIT