MCP (Model Context Protocol)
MCP is an open protocol from Anthropic that standardises how language models connect to external tools, data sources, and prompts — the USB-C of LLM integrations.
Model Context Protocol (MCP), introduced by Anthropic in late 2024 and widely adopted by 2026, defines a JSON-RPC interface between LLM clients (Claude Desktop, Cursor, Codex, custom agents) and MCP servers that expose tools, resources, and prompt templates. Instead of building a custom integration per tool per client, you write one MCP server per data source and any compliant client can use it. As of 2026, OpenAI, Anthropic, Google, Cursor, Cline, Continue, and Zed all implement MCP-compatible clients. Production stacks ship MCP servers for filesystem, GitHub, Postgres, Slack, internal APIs.
When to use mcp (model context protocol)
- Exposing internal data and tools to multiple LLM clients.
- Avoiding lock-in to a single LLM provider's tool format.
- Distributing capabilities as a reusable package.
Common mistakes
- Treating MCP as a wire format only — the prompt + tool surface design still matters.
- Shipping MCP servers without auth — same security concerns as any RPC endpoint.
FAQ
What is mcp (model context protocol)?
MCP is an open protocol from Anthropic that standardises how language models connect to external tools, data sources, and prompts — the USB-C of LLM integrations.
When should I use mcp (model context protocol)?
Exposing internal data and tools to multiple LLM clients. Avoiding lock-in to a single LLM provider's tool format. Distributing capabilities as a reusable package.
What are the most common mistakes with mcp (model context protocol)?
Treating MCP as a wire format only — the prompt + tool surface design still matters. Shipping MCP servers without auth — same security concerns as any RPC endpoint.
Related terms
- AI agent — An AI agent is a system where a language model autonomously plans and executes a sequence of tool calls to accomplish a goal.
- Function calling (tool use) — Function calling lets a language model emit a structured request to invoke a developer-defined tool, enabling reliable JSON output and agent workflows.
- Tool use (LLM) — Tool use is the umbrella term for any LLM mechanism that lets the model invoke external functions, APIs, or services — function calling, code interpreter, MCP servers, browser actions.
Sources
Last updated: 2026-06-01. Raw markdown: https://promtable.com/glossary/mcp.md.