ruijie89/mcp-server-android-emulator

MCP Ollama Server

A Model Context Protocol (MCP) server that provides integration with local Ollama LLM instances.

Features

• Text Generation: Generate text using any Ollama model
• Chat Interface: Have conversations with LLMs
• Model Management: List and switch between available models
• Server Status: Monitor Ollama server health
• Resource Access: Access server status and model information as resources

Prerequisites

1. Ollama: Install Ollama on your system

   # On macOS
   brew install ollama
   
   # On Linux
   curl -fsSL https://ollama.ai/install.sh | sh
   

2. Node.js: Version 18 or higher

3. Pull a model: Download at least one model in Ollama

   ollama pull llama2
   # or
   ollama pull codellama
   # or
   ollama pull mistral
   

Installation

1. Clone or create the project directory

2. Install dependencies:

   npm install
   

3. Build the TypeScript:

   npm run build
   

Usage

Starting the Server

npm start

The server communicates via stdio and is designed to be used with MCP-compatible clients.

Available Tools

1. generate_text

   • Generate text from a prompt
   • Parameters: prompt (required), model (optional), temperature (optional), max_tokens (optional)

2. chat

   • Have a conversation with the LLM
   • Parameters: messages (array of {role, content}), model (optional)

3. list_models

   • List all available Ollama models
   • No parameters required

Available Resources

1. ollama://status - Current server status and configuration
2. ollama://models - Detailed list of available models

Configuration

You can customize the server by modifying the constructor parameters in src/index.ts:

// Change Ollama URL (default: http://localhost:11434)
// Change default model (default: llama2)
const server = new OllamaMCPServer('http://localhost:11434', 'your-preferred-model');

Example Usage with MCP Client

// Generate text
{
  "method": "tools/call",
  "params": {
    "name": "generate_text",
    "arguments": {
      "prompt": "Explain quantum computing in simple terms",
      "model": "llama2",
      "temperature": 0.7
    }
  }
}

// Chat conversation
{
  "method": "tools/call",
  "params": {
    "name": "chat",
    "arguments": {
      "messages": [
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "What is the capital of France?"}
      ],
      "model": "llama2"
    }
  }
}

Development

• Build: npm run build
• Watch mode: npm run watch
• Development: npm run dev

Troubleshooting

1. "Connection refused" errors:

   • Make sure Ollama is running: ollama serve
   • Check if Ollama is accessible at http://localhost:11434

2. "Model not found" errors:

   • List available models: ollama list
   • Pull a model if needed: ollama pull llama2

3. Permission errors:

   • Make sure the built file is executable: chmod +x dist/index.js

Architecture

The server implements the MCP protocol and acts as a bridge between MCP clients and your local Ollama instance. It provides:

• Tools: Interactive functions that can be called by MCP clients
• Resources: Static or dynamic content that can be read by clients
• Stdio Transport: Communication via standard input/output for integration with MCP clients

License

MIT