Agentic

MCP.

Model Context Protocol. Anthropic's standard for AI agents to talk to external tools (databases, APIs, files). "USB-C for AI."

MCP (Model Context Protocol) is an open protocol that lets an LLM access the outside world — files, databases, APIs, tools — through a standard interface. Announced by Anthropic in late 2024, MCP is being positioned as the "USB-C of the AI ecosystem." MCP uses a client-server architecture: MCP servers expose capabilities (tools, resources, prompts), while LLM applications (Claude Desktop, Cursor, etc.) connect as MCP clients. Writing one MCP server lets you connect Slack, GitHub, Notion, and similar services to any AI in a single integration.

Model Context Protocol. AI agent'larının dış araçlarla (database, API, dosya) standart şekilde konuşmasını sağlayan Anthropic protokolü. "AI için USB-C."

MCP (Model Context Protocol), bir LLM'in dış dünyaya — dosyalara, veritabanlarına, API'lere, araçlara — standart bir arayüz üzerinden erişmesini sağlayan açık protokoldür. Anthropic tarafından 2024 sonunda duyurulan MCP, sektör için "AI ekosisteminin USB-C'si" olarak konumlandırılır. MCP istemci-sunucu mimarisi kullanır: MCP server'lar yetenek (tool, resource, prompt) sağlar; LLM uygulamaları (Claude Desktop, Cursor, vb.) MCP client olarak bağlanır. Bir MCP server yazmak Slack, GitHub, Notion gibi servisleri herhangi bir AI'a bir kez yazmakla bağlamak demektir.

Examples

claude mcp add filesystem npx @modelcontextprotocol/server-filesystem ~/Documents
claude mcp add postgres npx @modelcontextprotocol/server-postgres

Related terms.

001

Tool Use

Agentic

When the model decides to invoke a function instead of generating text. Same as "function calling." The core of agentic behavior.

002

Claude Code

Tools

Anthropic's terminal-based AI coding agent. Reads files, edits, runs commands, connects to external tools via MCP.

003

Agentic

When a model goes beyond replying — planning, calling tools, and acting toward a goal. Opposite of a passive chatbot.