modelcontextprotocol/quickstart-resources

A repository of servers and clients from the Model Context Protocol tutorials

What it solves

Model Context Protocol (MCP) is a standard that lets AI assistants like Claude call external tools. This repository provides minimal, working example servers and clients in five programming languages, so developers can learn how to build MCP-enabled applications without guessing at the details.

How it works

There are two kinds of examples:

  • Weather server (MCP server): A simple server that exposes two tools — get_forecast and get_alerts — backed by the US National Weather Service API. All servers communicate over stdio and expose the same two tools.
  • Chatbot (MCP client): An LLM-powered chatbot that launches a server, lists its tools, and runs an interactive chat loop in which Claude can call those tools. Clients need an ANTHROPIC_API_KEY, but without one they still connect and list the server's tools, so you can verify the MCP wiring without credentials.

Each example lives in its own directory with a README covering prerequisites, setup, and how to run it. You can mix and match across languages — for example, the Python client can drive the TypeScript server. The tests directory contains smoke tests that run in CI, verifying every server starts and answers MCP requests, and every client can connect to a mock server and list tools.

Who it's for

This is for developers who are learning the Model Context Protocol, want to see practical examples in a language they already use, or are looking for minimal, well-tested starting points to build their own MCP servers and clients. The examples intentionally stay minimal — they exist to teach the protocol, not to be production services.

Highlights

  • Example servers and clients in five languages: Python, TypeScript, Go, Rust, and Ruby
  • All servers expose the same two weather tools, so you can compare implementations directly
  • Clients are interactive chatbots that let Claude call the server's tools
  • Cross-language interoperability — clients and servers can be mixed and matched
  • Automated smoke tests run on every pull request, keeping examples reliable
  • Clear per-example READMEs with setup and run instructions

Related

  • Project
    JMAN730/opendroid

    OpenDroid is an open-source autonomous AI agent for Android that can understand complex voice or text commands, break them into step-by-step plans, and execute them across apps, system settings, and messaging—all while verifying results and adapting on failures.

  • Project
    i365dev/free4chat

    Free4Chat is an experimental, ephemeral collaboration platform that allows humans and AI agents to work together in temporary rooms without requiring accounts or permanent workspaces.

  • Project
    vercel-labs/workflow-builder-template

    A template for building AI-driven workflow automation platforms, featuring a visual drag-and-drop builder that generates executable TypeScript code.

  • Project
    mmalmi/nostr-vpn

    nostr-vpn is a Tailscale-style private mesh VPN using Nostr identities for enrollment and a FIPS-based data plane, with an experimental Bitcoin-paid exit-node marketplace. It offers a CLI, daemon, and native apps across desktop and mobile, plus server packages for StartOS and Umbrel. The fork claims ~3× throughput over the original FIPS implementation. Enrollment requires explicit admin approval; paid exits use Cashu channels with per-byte pricing.