In a move that bridges the gap between retro computing and modern AI agents, developer Andreas Wendorf (Akustikrausch) has released Amigent, a tool that enables AI agents like Claude Code and Codex to fully control a physical Commodore Amiga. The project, which appeared on Hacker News on November 4, 2026, allows users to automate tasks such as installing software, updating programs, and navigating the Workbench interface using natural language commands. Unlike typical remote desktop solutions, Amigent is specifically engineered to function over extremely slow network connections, including parallel port interfaces that deliver only about 50 KB/s.

Optimized for Legacy Hardware and Bandwidth

The core technical achievement of Amigent lies in its ability to maintain responsiveness on hardware that predates the internet age. The software is tuned for slow Amigas and slow networks, ensuring that a 7 MHz 68000 processor can transmit its entire screen state in under one second. By utilizing chipset screens sent as bitplanes and graphics-card screens as small palette tiles, the system minimizes data overhead. It only transmits changed tiles and calculates differences from the current state, allowing for near-instantaneous feedback even on a plipbox connection where raw pixel data would be prohibitively slow.

Hybrid Control for Humans and AI

Amigent supports simultaneous operation for both human users and AI agents, offering a remote desktop experience comparable to TeamViewer or VNC. The system prioritizes human input, processing mouse moves and keystrokes immediately even while screen updates are in transit. For AI-driven workflows, the agent can execute complex sequences like copying files or opening applications, while the user watches every click in real-time and can seize control at any moment. The setup process is designed to be agent-friendly, with a single prompt instructing the AI to install the necessary Python packages and register the MCP server, automating the network discovery and configuration on the host PC or Mac.

Key Takeaways

  • Amigent enables AI agents like Claude Code to perform autonomous tasks on real Commodore Amiga hardware.
  • The software is optimized for low-bandwidth environments, achieving full screen updates in under 1 second on a 7 MHz 68000 CPU.
  • It supports hybrid control, allowing humans to monitor and override AI actions in real-time over slow plipbox connections.
  • Installation is streamlined for AI agents, requiring only a single command prompt to set up the MCP server and remote viewer.

The Bottom Line

Amigent proves that AI agent utility isn't limited to cloud-native environments; with clever engineering, we can drive 40-year-old silicon with modern intelligence, effectively turning the Amiga into a controllable, autonomous peripheral.