mcp-function-calling-adapter

# mcp-function-calling-adapter English | [日本語](./README.ja.md) [](https://badge.fury.io/js/mcp-function-calling-adapter) [](https://opensource.org/licenses/MIT) [](https://github.com/d-kimuson/mcp-function-calling-adapter/actions) This library provides a function calling adapter for the Model Context Protocol (MCP). You can use standardized MCP Server implementations directly with function calling through this library. ## Installation ```bash pnpm add mcp-function-calling-adapter ``` ## Usage Here's an example using it with the OpenAI API: ```typescript import { McpFunctionCallingAdapter } from "mcp-function-calling-adapter" const adapter = new McpFunctionCallingAdapter("example", { "sequential-thinking": { command: "npx", args: ["-y", "@modelcontextprotocol/server-sequential-thinking"], }, }) const openai = new OpenAI({ apiKey: "your api key here", }) try { await adapter.startServers() const messages = [ { role: "user", content: "What's the weather like in Paris today?", }, ] const completion = await openai.chat.completions.create({ model: "gpt-4o-mini", messages, tools: adapter.getTools().map((tool) => ({ type: "function", function: { name: tool.name, description: tool.description ?? "", parameters: tool.inputSchema, }, })), tool_choice: "auto", }) // Call tool const toolCall = completion.choices.at(0)?.message.tool_calls?.at(0) if (toolCall && adapter.isRegisteredTool(toolCall.function.name)) { const response = await adapter.executeTool( toolCall.function.name, JSON.parse(toolCall.function.arguments) ) messages.push(completion.choices[0].message) messages.push({ role: "tool", tool_call_id: toolCall.id, content: response.content.toString(), }) } } finally { await adapter.clean() } ``` ## API ### `startServers()` Starts the MCP servers and loads available tools. ### `clean()` Disconnects from the MCP servers. ### `getTools()` Returns a list of available tools. The response includes the tool name, description, and input JSONSchema, which can be directly used for function calling schema definitions. ### `isRegisteredTool(name: string)` Checks if a tool with the specified name is registered and available for use. - `name`: The name of the tool to check - Returns: `boolean` - `true` if the tool is registered, `false` otherwise ### `executeTool(name: string, args: Record<string, unknown>)` Executes a tool with the specified name and returns the response from the MCP Server. - `name`: The name of the tool to execute - `args`: Arguments to pass to the tool ## Contributing We welcome contributions! Please see our [Contributing Guidelines](CONTRIBUTING.md) for details on how to submit pull requests, report issues, and contribute to the project. ### Development 1. Clone the repository 2. Install dependencies: `pnpm install` 3. Run tests: `pnpm test` 4. Build: `pnpm build` ## Changelog See [CHANGELOG.md](CHANGELOG.md) for a list of changes and version history. ## License This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.