Summary

agtx is a terminal-native orchestration tool for autonomous, multi-session, spec-driven AI coding. It enables collaborative AI agents to work on tasks, integrating with or defining custom spec-driven development frameworks. The benefit is streamlined, isolated, and highly customizable AI-assisted software development directly within the terminal.

How It Works

The system features a terminal TUI with a Kanban workflow (Backlog, Planning, Running, Review, Done). It uses tmux for session/window isolation (per project/task) and Git worktrees for task separation. agtx orchestrates multiple AI coding agents (Claude Code, Codex, Gemini, Copilot), allowing phase-specific agents with automatic switching and preserved context. Its plugin system supports spec-driven frameworks or custom workflows, defining per-phase skills, prompts, and artifact tracking for autonomous execution.

Quick Start & Requirements

Installation: curl -fsSL https://raw.githubusercontent.com/fynnfluegge/agtx/main/install.sh | bash. Prerequisites include tmux, gh (GitHub CLI), and supported AI coding agents (Claude Code, Codex, Gemini, OpenCode, Copilot). Start by running agtx in a Git repository or agtx -g for a multi-project dashboard.

Highlighted Details

• Terminal-native Kanban board for AI coding agent session management.
• Git worktree and tmux isolation per task for distinct development environments.
• Multi-agent collaboration support, enabling phase-specific AI models.
• Extensible plugin system for spec-driven frameworks (GSD, Spec-kit) or custom workflows.
• Automatic session management, context preservation, and agent switching.
• Integrated PR workflow for AI-generated descriptions and direct PR creation.

Maintenance & Community

The README provides no specific details on maintainers, community channels, sponsorships, or a public roadmap.

Licensing & Compatibility

The README does not specify the project's license. This omission requires further investigation for compatibility, especially concerning commercial use or integration with closed-source projects.

Limitations & Caveats

Setup requires tmux, gh, and configured AI coding agents. Functionality depends heavily on the user's AI agent setup and plugin implementation. The absence of explicit licensing information is a notable caveat for adoption.