MCP OpenVision

Name: MCP OpenVision
Author: Nazruden

@Nazruden

Repository

About MCP OpenVision

MCP Server using OpenRouter models to get descriptions for images

#mcp-openvision #image-analysis #ai-assistant

Basic information

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "mcp-openvision": {
      "command": "npx",
      "args": [
        "-y",
        "@smithery/cli",
        "install",
        "@Nazruden/mcp-openvision",
        "--client",
        "claude"
      ]
    }
  }
}

Tools

No tools detected

We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.

Overview

What is MCP OpenVision?

MCP OpenVision is a Model Context Protocol (MCP) server that provides image analysis capabilities powered by OpenRouter vision models. It enables AI assistants to analyze images through a simple interface within the MCP ecosystem.

How to use MCP OpenVision?

Install via pip install mcp-openvision or uv pip install mcp-openvision, or use Smithery for automatic integration with Claude Desktop. Configure by setting the OPENROUTER_API_KEY environment variable (required) and optionally OPENROUTER_DEFAULT_MODEL. Add the server to your MCP configuration file (e.g., mcp.json for Claude Desktop or Cursor) with the command uvx mcp-openvision and your API key. Test with npx @modelcontextprotocol/inspector uvx mcp-openvision.

Key features of MCP OpenVision

Single image_analysis tool accepting images as base64, URL, or file path.
Supports customizable system prompt, model, temperature, and max_tokens.
Works with any OpenRouter model that supports vision capabilities.
Default model is qwen/qwen2.5-vl-32b-instruct:free.
Can use relative file paths with an optional project_root parameter.
Integrates seamlessly with Claude Desktop and Cursor via MCP configuration.

Use cases of MCP OpenVision

Analyze store shelf images to identify retail products and estimate price ranges.
Examine medical scans for abnormalities and possible diagnoses.
Extract numerical data from charts and identify trends over time.
Transcribe text from menus, signs, or documents preserving structure.
Provide detailed art historical analysis of paintings focusing on composition and technique.

FAQ from MCP OpenVision

What is required to use MCP OpenVision?

You need an OpenRouter API key and a Python environment (pip or uv). The server itself runs as an MCP tool and requires no additional services beyond OpenRouter.

What vision models are supported?

Any OpenRouter model that supports vision capabilities. The default is qwen/qwen2.5-vl-32b-instruct:free, but you can change it via the OPENROUTER_DEFAULT_MODEL environment variable or by passing the model parameter directly.

How can I provide images to the tool?

Images can be given as base64‑encoded strings, HTTP/HTTPS URLs, or local file paths (absolute or relative). When using relative paths, you can specify a project_root to resolve them against a specific directory.

How do I integrate MCP OpenVision with Claude Desktop?

Edit your MCP configuration file (e.g., ~/.cursor/mcp.json or ~/Library/Application Support/Claude/claude_desktop_config.json) and add the openvision server entry with command uvx mcp-openvision and your OPENROUTER_API_KEY environment variable.