Guide

What is an MCP terminal client? (and how to use one)

TL;DR: agentty is an MCP-capable terminal coding agent: connect it to MCP servers to give the model real tools — file systems, databases, APIs — from the command line.

The Model Context Protocol (MCP) is an open standard that lets AI models call external tools and data sources through a consistent interface. An MCP terminal client is a command-line agent that can connect to MCP servers and use those tools during a session.

agentty is exactly that — a native, terminal-first MCP client.

Why MCP matters

Without MCP, an agent is limited to whatever its host wired in. With MCP, you can plug in standardized servers that expose:

  • File systems and repositories
  • Databases and search indexes
  • Web APIs and internal services
  • Custom company tooling

The model can then call these tools mid-task.

Using MCP with agentty

  1. Install agentty:
curl -fsSL https://agentty.org/install.sh | sh
  1. Configure the MCP servers you want available.
  2. Run agentty — the model can now invoke those tools during a conversation, with calls sandboxed by default.

Why a terminal MCP client

  • Native + fast — a single binary, no runtime.
  • Sandboxed — tool/shell calls are isolated by default.
  • Composable — pair MCP tools with agentty's built-in retrieval.

See the MCP docs for setup details.

Install agentty in one line — a single static binary.

curl -fsSL https://agentty.org/install.sh | sh
Get startedAll guides →