MCP
MCP
Model Context Protocol: a standard way for an agent to talk to outside tools
✓ Building BlocksWhat is a MCP
MCP, short for Model Context Protocol, is a standard that lets an agent connect to an outside tool or data source the same way regardless of who built either side. Before MCP, every integration between a model and a piece of software was its own custom wiring. MCP replaces that with one protocol: a server exposes what it can do, and any compliant client can find and use it.
In practice this means a running process, an MCP server, sits between the agent and the real system, whether that is a database, a SaaS product, or a local tool. When a session starts, the harness connects to the configured servers and asks each one what it offers. The agent sees a list of callable tools, calls one when needed, and gets a structured result back into context.
MCP is not a file the harness reads once and forgets. It is a live connection that can be queried repeatedly during a session. It also is not a replacement for a Skill: a Skill teaches the agent how to do something with what it already has, while MCP gives the agent something new to reach for in the first place.
The server is a separate, running process. The agent never talks to the underlying system directly, it always goes through the server's declared tools, which is what makes MCP swappable across different tools and vendors.
Build your own agent skills, not just prompts.
We design, build, and ship the Claude Code skill libraries production teams actually run on.
Talk to AY Automate ↗