Agent-Manager是一个开源的终端用户界面工具，可在单个tmux会话中同时运行多个AI编程代理 [1]。该工具支持Claude Code、Codex、OpenCode和Grok Build等多个AI助手 [1]，为用户提供了统一的平台来管理和与这些编程助手交互。
工具提供了会话管理、实时状态检测、代码差异审查、评论反馈和会话分组等核心功能 [1]。Agent-Manager支持macOS、Linux和WSL2等跨平台运行 [1]，用户可通过brew install yoanwai/tap/agent-manager或go install github.com/YoanWai/agent-manager@latest等方式安装 [1]。配置文件在macOS上位于/Library/Application Support/agent-manager/config.toml，在Linux上则位于/.config/agent-manager/config.toml [1]。该工具默认轮询间隔设置为2秒 [1]，并提供包括classic、solarized dark、catppuccin mocha等9种配色方案供用户选择 [1]。
Agent-Manager is an open-source terminal user interface (TUI) tool that enables users to run multiple AI programming agents simultaneously within a single tmux session.[1] The tool supports Claude Code, Codex, OpenCode, and Grok Build, providing a unified platform for managing interactions with various AI coding assistants.[1]
The application is available across multiple platforms, including macOS, Linux, and WSL2, with installation options via Homebrew (brew install yoanwai/tap/agent-manager) or Go (go install github.com/YoanWai/agent-manager@latest).[1] Agent-Manager offers core functionality including session management, real-time status monitoring, code difference review, comment feedback, and session grouping capabilities.[1] Users can configure the tool through platform-specific configuration files located at ~/Library/Application Support/agent-manager/config.toml on macOS or ~/.config/agent-manager/config.toml on Linux, with a default polling interval of 2 seconds.[1] The interface supports nine color schemes, including classic, solarized dark, and catppuccin mocha options.[1]