A terminal multiplexer that thinks it's an IDE (and an AI agent)

What it does Rune is a native development environment written in Go that combines a terminal multiplexer, a text editor/IDE, and an AI agent orchestration layer. It renders through OpenGL or Metal depending on your platform, explicitly avoiding Electron. Your workspaces are organized into nine slots holding infinite terminals, tabs, and windows, and the built-in networking layer—powered by Headscale—encrypts traffic between your own machines so you can resume work from any peer.

The interesting bit The README lists both “automatic and manual programming” as first-class use cases, meaning the same multiplexer is meant for human fingers and for the rune-agent extension. Designing a single UI for both you and an autonomous agent is an unusual constraint, and it likely explains the fixed nine-slot workspace model.

Key highlights

• Native GPU rendering (Metal on macOS, OpenGL on Linux) with no Electron shell.
• End-to-end encrypted peer networking via Headscale, so all your instances form a private mesh.
• “Batteries included” language intelligence, with a published list of supported languages.
• Plugin system that installs extensions directly from Git repositories.
• Dual personality: human IDE and agent orchestration environment, with a separate rune-agent binary.

Caveats

• Several Makefile targets for release, packaging, and cloud deployment are explicitly tied to the vendor’s internal infrastructure and “are not expected to work outside that environment.”
• The README advertises a built-in agent and “IDE-grade skills,” but does not detail what those agent capabilities actually are or how they differ from standard language intelligence features.

Verdict If you already treat a terminal multiplexer, a text editor, and an AI coding assistant as a single stack, Rune tries to collapse them into one native window. Everyone else should probably wait for the agent features to prove themselves beyond the README.