AIhelpAI-MCP

# AI Help AI MCP Server This is a server based on the Model Context Protocol (MCP) that allows AI agents to request help or consultation from other configured large language models (LLMs) via the MCP protocol. ## Features - **Multi-model Support**: Built-in access to various popular LLMs (e.g., Grok, Gemini, Claude, GPT-4o). - **Dynamic Tool List**: The MCP server automatically lists all available AI assistants as separate tools (e.g., `ask_grok3`, `ask_gemini2_5pro`). - **Unified Interface**: Interact with backend AI models through a standard OpenAI compatible API format. - **Conversation History**: Maintains a simple memory of conversation history (the last 5 rounds) to provide context for each model. - **Environment Configuration**: Configure the backend API proxy's URL and Key through environment variables. ## System Requirements - Node.js v18.0.0 or higher You can verify your Node.js installation with the following command: ```bash node --version # Should display v18.0.0 or higher ``` ## Installation Steps 1. Clone or download the project code. 2. Navigate to the project root directory. 3. Install dependencies: ```bash npm install ``` 4. Build the project: ```bash npm run build ``` (Note: `npm install` usually triggers `npm run build` automatically) ## Configuration Instructions (Client) This MCP server itself **does not** read the `.env` file. You need to configure the following environment variables in the **client application that starts this MCP server** (e.g., your AI chat interface or MCP management tool): 1. **`API_URL`**: The **base URL** pointing to your OpenAI compatible API proxy server (e.g., `http://localhost:3000`). The server code will automatically append the `/v1/chat/completions` path. 2. **`API_KEY`**: The key used to access your API proxy (e.g., `sk-xxxxxxxx`). ### Client Configuration Example (e.g., in VS Code MCP Plugin) In your client MCP server configuration interface (similar to the interface in your screenshot): - **Name**: Give this server a name, such as `Help AI`. - **Description**: A brief description, such as `Let AI seek help from other AIs`. - **Type**: Select `Standard Input/Output (stdio)`. - **Command**: Enter `node`. - **Arguments**: Enter the **absolute path** to the **compiled** `index.js` file of this project. For example: `actual_path\AIHelpAI-MCP\build\index.js` (please modify according to your actual path). - **Environment Variables**: - Add `API_URL`, with the value being your API proxy URL (e.g., `http://localhost:3000`). - Add `API_KEY`, with the value being your API key (e.g., `sk-xxxxxxxx`). After saving the configuration, the client should be able to start and connect to this MCP server. ## Usage After configuring and starting the server, AI agents connected to this MCP server will see a series of available tools in the format `ask_<model_key>`, such as: - `ask_grok3` - `ask_gemini2_5pro` - `ask_claude3_7sonnet` - `ask_gpt4o` Each tool's description will explain the characteristics and areas of expertise of that AI assistant. When calling a tool, simply provide one parameter: - `prompt` (string, required): The question or instruction you want to pose to the AI assistant. The server will use the configured API URL and Key, along with the built-in model parameters and conversation history, to make a request to the corresponding model and return the response to the caller. ## Developer Tools You can use the MCP Inspector tool to test the server (if globally installed or using npx): ```bash # Ensure that API_URL and API_KEY environment variables are set npx @modelcontextprotocol/inspector build/index.js # Or, if installed globally mcp-inspector build/index.js ``` ## License Please refer to the license file in the project repository.