eMCP Server

Overview

eMCP is a lightweight, TypeScript‑based Model Context Protocol (MCP) server that extends the functionality of its predecessor LiteMCP. It addresses a common pain point for developers who need to expose AI‑ready tools, resources, and prompts while maintaining control over authentication and request handling. By bundling a built‑in authentication handler and a flexible middleware stack, eMCP lets teams secure their MCP endpoints without having to write boilerplate code or integrate separate authentication services.

The core value of eMCP lies in its “drop‑in replacement” promise. Existing projects that already depend on LiteMCP can upgrade with minimal friction: all original features remain intact, and the new capabilities are optional. Developers simply add an callback or register middleware functions, and the server will enforce those policies for every incoming request. This makes eMCP an attractive choice for internal tooling, rapid prototyping, or production services where consistent security and observability are required.

Key features include:

• Built‑in authentication: A pluggable handler that receives the raw request and returns a boolean or error, allowing custom logic such as API key checks, OAuth validation, or JWT verification.
• Layered middleware: Inspired by frameworks like Hono and Express, middleware runs in registration order before the MCP handler and reverses after the response is generated. This enables logging, timing, rate‑limiting, or request transformation without touching the core tool logic.
• Full LiteMCP compatibility: Tools, resources, prompts, and sampling work exactly as before, so developers can continue to use familiar patterns for defining tools or extending the protocol.
• TypeScript safety: All request and response types are strongly typed, reducing runtime errors and improving IDE support.

Typical use cases include:

• Secure AI assistants: Deploy a private MCP server that only allows authenticated users to invoke computational tools, ensuring sensitive data is never exposed.
• Observability and analytics: Wrap the request pipeline in middleware that logs latency, counts usage per endpoint, or integrates with monitoring systems.
• Custom request shaping: Pre‑process incoming arguments (e.g., normalizing dates or sanitizing strings) before they reach the tool implementation.
• Gradual migration: Add authentication to an existing LiteMCP deployment without refactoring the tool definitions.

Integration into AI workflows is straightforward. A Claude or other MCP‑aware assistant simply points to the eMCP endpoint; the server handles authentication, applies any middleware transformations, and routes the request to the appropriate tool. The assistant receives a standard MCP response, allowing developers to focus on business logic rather than protocol plumbing.

eMCP’s standout advantage is its minimal friction upgrade path coupled with powerful extensibility. By keeping the base API identical to LiteMCP and layering optional security and middleware, it empowers teams to add robust safeguards and custom behaviors while preserving the simplicity that made MCP attractive in the first place.