@mcpjam/sdk

The official MCPJam SDK provides utilities for building, testing, and developing MCP clients and servers. Built on top of the @modelcontextprotocol/sdk, it offers high-level abstractions and tools to accelerate MCP development.

Installation

npm install @mcpjam/sdk

Key Features

MCPClientManager

The primary utility in the SDK is MCPClientManager, a powerful client manager for connecting to and interacting with MCP servers:

• Multi-server support - Manage multiple MCP server connections simultaneously
• All transports - STDIO, HTTP/SSE, and Streamable HTTP support
• Lifecycle management - Automatic connection handling and cleanup
• Tools, resources, prompts - Full MCP protocol support including elicitation
• Agent framework integration - Built-in adapters for Vercel AI SDK and other popular libraries
• OAuth & authentication - Bearer token and custom header support

Use Cases

The SDK is designed for:

• Building AI agents - Connect agents to MCP servers for tool access
• Creating MCP clients - Build custom clients with full protocol support
• Testing MCP servers - Write unit tests and E2E tests for your servers
• LLM applications - Add MCP support to chat applications and AI workflows

Quick Start

import { MCPClientManager } from "@mcpjam/sdk";

// Initialize with server configurations
const manager = new MCPClientManager({
  // STDIO server
  filesystem: {
    command: "npx",
    args: ["-y", "@modelcontextprotocol/server-filesystem", "/tmp"],
  },
  // HTTP/SSE server with authentication
  asana: {
    url: new URL("https://mcp.asana.com/sse"),
    requestInit: {
      headers: {
        Authorization: "Bearer YOUR_TOKEN",
      },
    },
  },
});

// List and execute tools
const tools = await manager.getTools(["filesystem"]);
const result = await manager.executeTool("filesystem", "read_file", {
  path: "/tmp/example.txt",
});

// Integrate with Vercel AI SDK
import { generateText } from "ai";
import { openai } from "@ai-sdk/openai";

const response = await generateText({
  model: openai("gpt-4o-mini"),
  tools: manager.getToolsForAiSdk(),
  messages: [{ role: "user", content: "List files in /tmp" }],
});

Documentation

For detailed documentation on MCPClientManager including:

• Connection configuration (STDIO, HTTP/SSE)
• Tool execution and resource management
• Elicitation handling
• Agent framework integrations
• API reference

See the MCPClientManager README.

Development

Building Locally

Build the entire SDK workspace:

npm run build

This compiles all sub-packages including mcp-client-manager and generates distributable bundles.

Development Mode

Watch for changes and rebuild automatically:

npm run dev

Resources

• 💬 Discord: Join the MCPJam Community
• 📖 MCP Protocol: Model Context Protocol Documentation
• 🔧 GitHub: MCPJam Inspector Repository

Contributing

We welcome contributions! The SDK is part of the MCPJam Inspector monorepo. Please see our Contributing Guide for guidelines.

License

Apache License 2.0 - see the LICENSE file for details.

Built with ❤️ for the MCP community • 🌐 MCPJam.com