MCP in Practice

Anthropic's Model Context Protocol (MCP) is a system designed to make AI tools and platforms model-agnostic by defining servers for tools and clients for AI applications. An analysis of the MCP ecosystem, based on GitHub star counts for 2,874 servers, reveals significant concentration, with the top 10 servers accounting for nearly half of all developer attention. The dominant use cases fall into three main categories: Database & Search, Computer & Web Automation, and Software Engineering. Most MCP servers support both read and write operations, indicating a strong developer interest in creating AI agents that can act on context, not just consume it. The popularity of agentic browsing tools further suggests that developers are using MCP to bypass the limitations of existing APIs.