Toad provides a unified, terminal-based interface for running AI coding agents, leveraging the ACP protocol. It aims to simplify the management and execution of various AI agents within a single, user-friendly TUI, benefiting developers and power users seeking streamlined AI integration in their workflow.

How It Works

The project utilizes the ACP (Agent Communication Protocol) to abstract away the complexities of interacting with different AI agents. This allows Toad to present a consistent, beautiful terminal UI for discovering, installing, and launching coding agents, offering a seamless experience without requiring users to manage individual agent interfaces or environments.

Quick Start & Requirements

Installation is straightforward via a curl script:

curl -fsSL batrachian.ai/install | sh

An alternative method involves installing uv first, then running:

uv tool install -U batrachian-toad --python 3.14

Toad is compatible with Linux and macOS; native Windows support is absent but functional via WSL. A recommended terminal for macOS is Ghostty for a full experience. Launch agents with toad or toad -a <agent_name>. A web server can be started with toad serve.

Highlighted Details

• Unified terminal UI for AI agents.
• Support for discovering, installing, and launching agents.
• Command-line interface (CLI) with project directory awareness and agent launch shortcuts.
• Optional web server mode for browser-based access.

Maintenance & Community

Developed by Will McGugan, Toad is under active development. Community interaction is encouraged via the Discussions tab on GitHub and the #toad channel on the Textualize Discord server. A roadmap includes features like UI for MCP servers and model selection exposure. Bug reporting follows a non-traditional process, requiring prior discussion approval.

Licensing & Compatibility

The provided README does not specify a software license.

Limitations & Caveats

Native Windows support is not provided, though it functions via WSL. The default macOS terminal offers a reduced experience compared to recommended alternatives like Ghostty. The issue reporting process requires pre-approval in discussions, potentially adding friction for bug submission.