LLM MCP Server

A Model Context Protocol (MCP) server that enables AI clients to query multiple Large Language Models (OpenAI GPT, Anthropic Claude, and Google Gemini) through a unified interface.

Prerequisites

• Node.js (v16 or higher recommended)
• npm (comes with Node.js)
• API keys for the LLM providers you want to use:
  • OpenAI API key (for GPT models)
  • Anthropic API key (for Claude models)
  • Google API key (for Gemini models)
  • OpenRouter API key (for Qwen models)

Installation

Clone the repository and install dependencies:

npm install

Usage

This MCP server provides five tools for querying different AI models:

• ask_gpt - Query OpenAI GPT models (default: gpt-4o-2024-11-20)
• ask_claude - Query Anthropic Claude models (default: claude-4-sonnet)
• ask_gemini - Query Google Gemini models (default: gemini-2.5-flash)
• ask_qwen - Query Qwen models via OpenRouter (default: qwen/qwq-32b-preview)
• ask_deepseek - Query DeepSeek models via OpenRouter (default: deepseek/deepseek-chat-v3-0324)

Available Tools

Each tool accepts the following parameters:

Recommended Models

OpenAI GPT (ask_gpt):

• gpt-4o-2024-11-20 (default) - Latest GPT-4 model
• o3 - Reasoning model for complex problems
• o4-mini - Faster reasoning model

Anthropic Claude (ask_claude):

• claude-sonnet-4-20250514 (default) - Balanced performance
• claude-3-5-haiku-20241022 - Speed optimized

Google Gemini (ask_gemini):

• gemini-2.5-flash (default) - Price/performance balance
• gemini-2.5-pro - Complex problems
• gemini-2.5-flash-lite - Speed/cost optimized

Qwen via OpenRouter (ask_qwen):

• qwen/qwq-32b-preview (default) - Reasoning tasks
• qwen/qwen-2.5-72b-instruct - General tasks

DeepSeek via OpenRouter (ask_deepseek):

• deepseek/deepseek-chat-v3-0324 (default) - Advanced reasoning and general tasks

Connecting to MCP Clients

This server is designed to work with MCP-compatible clients like Claude Desktop. Add it to your MCP client configuration to access the LLM querying tools.

{
    "ask-llm": {
        "command": "node",
        "args": [
            "path/to/llm-mcp/index.js"
        ],
        "env": {
            "OPENAI_API_KEY": "your-openai-key",
            "ANTHROPIC_API_KEY": "your-anthropic-key",
            "GOOGLE_API_KEY": "your-google-key",
            "OPENROUTER_API_KEY": "your-openrouter-key"
        }
    }
}

You only need API keys for the models you plan to use.

## Local LLM (per-request base_url)

The `ask_local_llm` and `list_local_models` tools support an optional per-request `base_url` parameter. This lets you point a single MCP server at different OpenAI-compatible local LLM servers without changing environment variables. Example tool input for `ask_local_llm`:

```json
{
    "question": "Summarize the README",
    "model": "llama3.2",
    "base_url": "http://localhost:1234/v1"
}
```

For `list_local_models`, pass `{ "base_url": "http://other-host:1234/v1" }` to list models from an alternate local server.

Contributing

Pull requests are welcome! For major changes, please open an issue first to discuss what you would like to change.

License

ISC