About MCP Serverless
MCP Serverless is a package for building serverless MCP servers and managing their tools. It includes an in-memory client-server setup and supports passing context with requests for client credentials.
Key features
- 01Register and manage tools
- 02Handle tool requests
- 03Create in-memory client-server connections
- 04Pass request context for credentials
Use cases
- 01Build a serverless MCP service
- 02Connect a client and service in memory
- 03Forward client credentials through request context
MCP support
Provides APIs to create MCP servers and in-memory client-server services, with examples using stdio transports.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEMCP Serverless
A serverless implementation for the Model Context Protocol (MCP) architecture that enables tool management through a clean interface.
Overview
This package provides a serverless implementation of the MCP server that allows you to:
- Register and manage tools
- Handle tool-related requests
- Create in-memory client-server connections
- Extend request with context to enable credential transmission from clients
Installation
npm install @tilfin/mcp-serverless
Usage
Tool Registration and create a client for service implements
import { createService, ToolManager } from '@tilfin/mcp-serverless';
// Create a tool manager
const toolManager = new ToolManager();
// Register tools
toolManager.registerTools([
{
name: 'calculator',
description: 'Performs basic arithmetic operations',
inputSchema: {
type: 'object',
properties: {
operation: { type: 'string' },
numbers: { type: 'array', items: { type: 'number' } }
},
required: ['operation', 'numbers']
},
toolFunction: async (params, ctx) => {
if (ctx.apiKey !== 'xyz') throw new Error('Invalid API Key');
let result;
if (params.operation === 'add') {
result = params.numbers.reduce((sum, n) => sum + n, 0);
}
return { result };
}
}
]);
// Create a serverless client
const client = createService(toolManager);
// List available tools
const toolsList = await client.listTools();
// Call a tool
try {
const result = await client.callTool({
name: 'calculator',
arguments: {
operation: 'add',
numbers: [1, 2, 3]
}
});
} catch (err) {
// raise Invalid API Key error
}
// Call a tool with context
const result = await client.callTool({
name: 'calculator',
arguments: {
operation: 'add',
numbers: [1, 2, 3]
},
ctx: { apiKey: 'xyz' }
});
API Reference
ToolManager class
Manages the registration and handling of tools.
Tool Interface
Tools must implement the following interface:
interface Tool {
name: string;
description: string;
inputSchema: ToolInput;
toolFunction: (args: CallToolRequestArguments, ctx: CallToolRequestContext) => Promise<CallToolResultContent>;
}
createServer(serverInfo, toolManager)
Creates an MCP server with the given tool manager.
createService(toolManager)
Creates an in-memory client-server setup for serverless operation.
Examples
Standard I/O Transport
The package includes examples for using the stdio transports for both client and server communication:
StdioClientTransport: Allows clients to communicate with servers over standard input/outputStdioServerTransport: Enables servers to handle requests through standard input/output
Check out the example implementation:
- Client: stdio_client.mjs
- Server: stdio_server.mjs
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.