mcp-bundler
Is the MCP configuration too complicated? You can easily share your own simplified setup!
claude mcp add --transport stdio wrtnlabs-mcp-bundler bun --watch /path/to/figma-mcp/src/index.ts \ --env PORT="RequiredEnv" \ --env FIGMA_PERSONAL_ACCESS_TOKEN="RequiredEnv"
How to use
MCP Bundler lets you consolidate multiple MCP server configurations into a single cohesive server bundle. In practice, you describe each MCP-backed service you want to expose (for example a Figma-based MCP, a Notion API MCP, or another custom server) under mcpServers. Each entry specifies how to launch that particular MCP server (the command and arguments) and any environment variables it needs. Once bundled, you can run the whole server bundle and have all included MCPs start together, share settings, and be orchestrated as a single or composable server instance. The tools exposed in the bundle include support for standard MCP modes like stdio, SSE, and InMemory, depending on how you configure and invoke the server core. The configuration demonstrates mixing a local script-based MCP (bun with a TypeScript entry) and a published MCP server (Notion API) that can be started via npx.
How to install
Prerequisites:
- Node.js and npm (or pnpm/yarn if you prefer).
- Basic familiarity with MCP server concepts and the bundler package.
Install the MCP Bundler package locally (example with npm):
npm install --save-dev @wrtnlabs/mcp-bundler
Create a configuration file (e.g., mcp.config.json) or integrate the config into your startup script, using the examples from the README as a template.
Run the bundled MCP server:
node path/to/your/bundle-entry.js
If you prefer using npx to run the bundler with your config, you can also do:
npx @wrtnlabs/mcp-bundler -c path/to/mcp.config.json
Note: The exact commands for each MCP server inside mcpServers depend on what you need to run (bun, npx, node scripts, etc.). The examples in the README demonstrate how to mix a local bun-based TS entry and a Notion MCP via npx.
Additional notes
Tips and caveats:
- Ensure required environment variables are marked as RequiredEnv in your mcp config to enforce they’re provided at runtime.
- For local TypeScript entries (e.g., /path/to/figma-mcp/src/index.ts), you may be using bun to execute TS directly; confirm your runtime supports TS execution in your environment.
- If any MCP servers require ports, ensure those ports are not in conflict with other processes and are exposed as needed.
- You can extend mcpServers with additional entries following the same shape (command, args, env).
- When debugging, run individual MCP entries in isolation to verify they load correctly before bundling them, which helps isolate config mistakes quickly.
Related MCP Servers
mcp-router
A Unified MCP Server Management App (MCP Manager).
agent
A unified Model Context Protocol server implementation that aggregates multiple MCP servers into one.
pluggedin-app
The Crossroads for AI Data Exchanges. A unified, self-hostable web interface for discovering, configuring, and managing Model Context Protocol (MCP) servers—bringing together AI tools, workspaces, prompts, and logs from multiple MCP sources (Claude, Cursor, etc.) under one roof.
yamcp
Organize your MCP servers in local workspaces, share them as Yet-Another-MCP through a single command
mcp -zotero-dev
Give your AI assistant superpowers for Zotero plugin development. 25 tools for screenshots, DOM inspection, JavaScript execution, build integration, and debugging via Model Context Protocol.
mcp-more
A modern desktop application for managing Model Context Protocol (MCP) servers.