Skip to main content
Back

Will McGugan's Toad Unifies Claude Code and Gemini CLI in One Terminal

Toad is a TUI that unifies Claude Code, Gemini CLI, OpenHands, and 15+ other coding agents under one polished terminal interface

Repo, Full Access
  • Toad is a Python TUI (terminal user interface) that acts as a unified front-end for 18+ AI coding agents including Claude Code, Gemini CLI, Codex, and OpenHands.
  • It integrates a real shell with full color, interactivity, and tab completion — the only terminal agent UI to do this at launch.
  • Multiple agents from different providers can run concurrently; switch between them with ctrl+s and resume past sessions with ctrl+r.
  • Built on the Agent Client Protocol (ACP), an open standard backed by Zed, JetBrains, Anthropic, OpenAI, and Google.
  • Available now on Linux and macOS (WSL on Windows); install via curl -fsSL batrachian.ai/install | sh. Licensed under AGPL with commercial options.
  • Created by Will McGugan (author of Rich and Textual) after his terminal startup Textualize shut down; the repo has 3,300+ stars.

Toad is a terminal UI for coding agents built by Will McGugan, the creator of Rich and Textual. The idea is simple but overdue: instead of juggling separate CLIs for Claude Code, Gemini CLI, Codex, OpenHands, and whatever comes next, you get one consistent, keyboard-driven interface that hosts them all. It has crossed 3,300 GitHub stars and is actively maintained.

The backstory matters. McGugan was the CEO of Textualize, a startup promoting rich terminal applications, which didn't make it as a company. When he finally tried Claude Code and Gemini CLI, he was more interested in the terminal interface than the AI, and he was not impressed. Both projects suffer from jank and glitches that Textualize had solved years ago. So he built Toad.

One front-end, every agent

Toad provides a front-end for AI tools such as OpenHands, Claude Code, Gemini CLI, and other ACP-enabled agents, allowing them to run within a single terminal application rather than separate CLIs. At the time of writing, it supports 12+ agent CLIs. The glue holding this together is the Agent Client Protocol (ACP).

ACP is an open standard that abstracts the events and outputs of AI agents and provides a common interface for editors to interact with them. Think of it as MCP but for agent-to-client (UI) communication. ACP and MCP address different layers of the agentic stack and are complementary rather than competing: a single coding agent can speak ACP to its host editor while separately using MCP to call out to external tools.