Model Context Protocol (MCP) Resource Overview

Explore the essential MCP documentation – from connecting local and remote servers, building servers and clients, to SDKs and security guidance – all curated in concise summaries.

Connect to local MCP servers

This guide shows how to configure Claude Desktop (and other clients) to run a local MCP Filesystem server, granting AI controlled access to your computer’s files and folders while requiring explicit user approval for every action.

Connect to remote MCP servers

Learn how to attach Claude Desktop (or any MCP‑compatible client) to remote MCP servers via Custom Connectors, enabling AI assistants to call external tools and services over the internet with secure authentication and permission controls.

Build an MCP server

Step‑by‑step instructions for creating a weather‑focused MCP server in multiple languages (Python, TypeScript, Java, Kotlin, C#, Rust). The tutorial covers tool definition, logging best practices, transport setup, and integration with Claude Desktop.

Build an MCP client

Detailed walkthrough for constructing a chatbot client that talks to MCP servers, covering environment setup, API key handling, server connection, tool execution, interactive REPL, and error‑handling patterns across Python, TypeScript, Java, Kotlin, and C#.

Understand MCP SDKs

Overview of the official MCP SDKs available for TypeScript, Python, Java, C#, Go, Kotlin, Swift, Rust, Ruby, and PHP, highlighting installation, core capabilities, and links to quick‑start resources for building compliant servers and clients.

MCP Security Overview

Summarizes MCP’s authorization model based on OAuth 2.1, token validation, audience checks, and best‑practice recommendations for consent handling, input validation, HTTPS enforcement, and mitigation of common attacks like SSRF and token leakage.