Model Context Protocol

Model Context Protocol (MCP) is an open standard introduced by Anthropic that defines how AI language models communicate with external data sources and tools. Think of it as USB-C for AI — one plug that works everywhere, letting any compatible agent connect to any compatible server without custom glue code.

Before MCP, connecting an AI agent to a database or API meant writing brittle, one-off integrations for every combination of model and tool. MCP replaces that with a single protocol: the agent speaks MCP, the server speaks MCP, and everything just works. This dramatically reduces setup time and makes integrations portable across Claude Code, Cursor, Windsurf, and any other MCP-compatible client.

An MCP server exposes tools (actions the agent can invoke), resources (data the agent can read), and prompts (reusable templates). The agent decides when and how to use them — the server just provides the capability.

Skills tell your agent how to behave — coding style, workflow rules, and task-specific instructions. MCP servers tell your agent what it can do — query a database, open a pull request, or send a Slack message.

The combination is powerful: a skill might instruct your agent to always run tests before committing, while the GitHub MCP server gives it the tools to actually create the commit and open the PR. Skills define the policy; MCP provides the capability.