MCP

MCP — explained in more detail

The Model Context Protocol is an open standard introduced by Anthropic to unify how external data sources and tools are wired into AI models. Instead of building a custom interface for every integration, MCP servers and MCP clients speak a shared protocol. Databases, APIs, file systems and issue trackers are exposed as tool servers.

Practical relevance

MCP servers are interchangeable across AI clients. A server that exposes access to an issue tracker, for example, can be used equally from Claude Desktop, Cursor or Claude Code. That lowers integration cost and shortens the path from “tool exists” to “tool is usable”.

Distinction from similar terms

MCP is not an AI model and not an agent — it is the underlying connectivity layer. Compared to a single vendor’s function-calling, MCP defines a cross-vendor interface: the same tool server works with different models and clients.

More on this topic

See everything in one place:Agents→

News · 05 Jun 2026Anthropic: Claude Writes Over 80% of Its Own Code — and Calls for a Pause ButtonAnthropic discloses the numbers: Claude writes >80% of its production code. At the same time, the company argues for the option to pause frontier development.News · 22 Jun 2026The Ralph Loop Goes Mainstream in 2026 — and Brings a Token-Burn ProblemAutonomous coding agents in an endless loop: what Ralph Loop and Loop Engineering really are — and why the bill explodes without hard stops.Encyclopedia · 13 Jun 2026Function Calling / Tool UseHow an LLM uses tools: define a tool as a schema, the model picks the function and arguments, the result returns to the chat — the basis of every agent.