Vision MCP Server

Local setup required. This server has to be cloned and prepared on your machine before you register it in Claude Code.
1

Set the server up locally

Run this once to clone and prepare the server before adding it to Claude Code.

Run in terminal
npm install -g @thenomadinorbit/vision-mcp-server
2

Register it in Claude Code

After the local setup is done, run this command to point Claude Code at the built server.

Run in terminal
claude mcp add -e "OPENROUTER_API_KEY=${OPENROUTER_API_KEY}" vision-mcp -- node "<FULL_PATH_TO_VISION_MCP_SERVER>/dist/index.js"

Replace <FULL_PATH_TO_VISION_MCP_SERVER>/dist/index.js with the actual folder you prepared in step 1.

Required:OPENROUTER_API_KEY+ 1 optional
README.md

Adds vision capabilities to any AI model through OpenRouter

Vision MCP Server

Ever wanted to use a model like GLM-4.6 or other great AI models that just don't have vision capabilities? This MCP server solves that problem by adding vision capabilities to any model through OpenRouter's vision models.

The Problem

Some really good AI models don't support vision. You're stuck choosing between your preferred model or vision capabilities. This server bridges that gap by providing seamless vision capabilities through OpenRouter's vision models.

The Solution

This MCP server provides a simple analyze_image tool that can:

  • Analyze images from URLs, file paths, or base64 data
  • Use any vision model available on OpenRouter (Claude 3.5 Sonnet, GPT-4 Vision, etc.)
  • Return detailed analysis results
  • Handle errors gracefully with proper validation

System Requirements

Before installing, make sure you have:

  • Node.js 18.0.0 or higher (recommended: Node.js 20+)
  • npm 8.0.0 or higher (comes with Node.js)

Check Your Versions

node --version    # Should show v18.0.0 or higher
npm --version     # Should show 8.0.0 or higher

Install/Update Node.js

If you need to install or update Node.js:

  1. Download from official site: nodejs.org (recommended for beginners)
  2. Using Node Version Manager (nvm):
    # Install nvm first, then:
    nvm install 20
    nvm use 20
    
  3. Using package managers:
    • macOS: brew install node
    • Windows: winget install OpenJS.NodeJS
    • Ubuntu/Debian: sudo apt install nodejs npm

Important: This server is written in TypeScript and uses dependencies (like node-fetch v3) that require Node.js 18+. Older versions (like Node.js 16 or below) will not work.

Quick Start

Step 1: Get Your OpenRouter API Key

  1. Go to OpenRouter
  2. Sign up or log in to your account
  3. Navigate to "Keys" in your dashboard
  4. Click "Create Key"
  5. Copy your API key (starts with sk-or-v1-...)
  6. Keep this key safe - you'll need it in Step 3

Step 2: Install the MCP Server

Option A: Install from npm (Recommended)
npm install -g @thenomadinorbit/vision-mcp-server

Success! The package is now globally available as vision-mcp command.

Option B: Install from Source (Development)
git clone https://github.com/TheNomadInOrbit/vision-mcp-server.git
cd vision-mcp-server
npm install
npm run build
npm install -g .

Note: Use this method if you want to modify the source code or contribute to the project.

Step 3: Configure Your MCP Client

Add this server configuration to your MCP client:

{
  "mcpServers": {
    "vision-analyzer": {
      "command": "vision-mcp",
      "type": "stdio",
      "timeout": 60,
      "disabled": false,
      "autoApprove": [],
      "env": {
        "OPENROUTER_API_KEY": "your_api_key_here",
        "OPENROUTER_MODEL": "anthropic/claude-3-5-sonnet"
      }
    }
  }
}

Step 4: Test Your Installation

Important: The vision-mcp command requires an OpenRouter API key to run. You cannot test it directly without configuration.

Quick Test (with your API key):
OPENROUTER_API_KEY="your_api_key_here" vision-mcp --help

You should see the server start up with logs like:

Application initialized successfully
Starting Vision MCP Server...
MCP server started successfully
Vision MCP Server is running on stdio

Press Ctrl+C to stop the test.

What happens if you run `vision-mcp` without the API key?
vision-mcp

You'll get this error (this is normal and expected):

Error: OPENROUTER_API_KEY environment variable is required

This means the installation worked! The server is just protecting you from running without proper configuration.

Verify Installation Status:
# Check if the command is available
which vision-mcp

# Check if the package is installed
npm list -g @thenomadinorbit/vision-mcp-server

🔧 Configuration Options

Basic Configuration

  • "vision-analyzer" - Server name (you can change this to anything you like)
  • "command": "vision-mcp" - Required: The global command to run the server
  • "type": "stdio" - Required: Communication protocol for MCP
  • "timeout": 60 - Optional: Timeout in seconds (default: 60)
  • "disabled": false - Optional: Set to true to disable the server

Auto-Approve Settings

Configure which tools can run without asking for permission:

"autoApprove": []

Options:

  • [] (empty) - Requires approval for all tools (safest)
  • ["list_models"] - Auto-approve listing available models only
  • ["analyze_image"] - Auto-approve vision analysis (convenient but less safe)
  • ["analyze_image", "list_models"] - Auto-approve all tools (most convenient)

Mod

Tools (1)

analyze_imageAnalyzes images from URLs, file paths, or base64 data using OpenRouter vision models.

Environment Variables

OPENROUTER_API_KEYrequiredYour API key from OpenRouter
OPENROUTER_MODELThe specific vision model to use from OpenRouter

Configuration

claude_desktop_config.json
{
  "mcpServers": {
    "vision-analyzer": {
      "command": "vision-mcp",
      "type": "stdio",
      "env": {
        "OPENROUTER_API_KEY": "your_api_key_here",
        "OPENROUTER_MODEL": "anthropic/claude-3-5-sonnet"
      }
    }
  }
}

Try it

Analyze the image at this URL and tell me what text is written on the sign.
Look at this local file path and describe the layout of the UI elements.
Can you identify the objects in this image and provide a JSON list of them?
Analyze the provided image and explain the chart data shown.

Frequently Asked Questions

What are the key features of Vision MCP Server?

Adds vision capabilities to models that lack native image support. Supports image analysis via URLs, local file paths, or base64 data. Integrates with any vision model available on OpenRouter. Provides detailed analysis results with graceful error handling.

What can I use Vision MCP Server for?

Extracting text from images using models that don't natively support vision. Analyzing UI screenshots for development feedback. Interpreting charts and graphs from local image files. Identifying objects or scenes in images for automated documentation.

How do I install Vision MCP Server?

Install Vision MCP Server by running: npm install -g @thenomadinorbit/vision-mcp-server

What MCP clients work with Vision MCP Server?

Vision MCP Server works with any MCP-compatible client including Claude Desktop, Claude Code, Cursor, and other editors with MCP support.

Turn this server into reusable context

Keep Vision MCP Server docs, env vars, and workflow notes in Conare so your agent carries them across sessions.

Need the old visual installer? Open Conare IDE.
Open Conare