What happens when you try to build one interface for every AI agent you're running? Lukan Labs thinks the answer is a single Rust binary that combines a terminal, browser automation, messaging integrations, and multi-agent orchestration into one workspace.
Lukan is an open-source (MIT licensed) AI workstation written in roughly 49,000 lines of Rust. It ships as a single binary with no runtime dependencies and supports 10 LLM providers out of the box: Anthropic, OpenAI, GitHub Copilot, Gemini, Fireworks, Nebius, Ollama, and several others. Bring your own API keys, no vendor lock-in.
What It Actually Does
The core pitch is managing multiple AI agents from one place. You can run several agent tabs simultaneously, each working on different tasks, with session checkpoints that let you rewind conversations. When tmux (a terminal multiplexer that lets you run multiple terminal sessions in one window) is available, Lukan uses it for persistent sessions that survive browser crashes and reconnects. Without tmux, it falls back to standard terminal sessions.
Beyond the terminal, Lukan includes browser automation via Chrome DevTools Protocol, plugin-based messaging (WhatsApp, Telegram, Slack, Discord, email), and a DAG-based pipeline system for chaining agents together in multi-step workflows. There's also a background worker feature for scheduling autonomous tasks on a cron schedule.
Security-wise, it uses sandboxed execution with OS-level isolation and offers end-to-end encrypted remote access using X25519 key exchange.
Early Days, Big Surface Area
The ambition here is wide. Lukan is trying to be a coding assistant, a browser agent, a messaging hub, and a workflow orchestrator all at once. That's a lot of surface area for a project that just appeared on GitHub. The four interface options (terminal UI, web UI, desktop app, CLI) suggest the team is casting a wide net to see what sticks.
For developers juggling Claude Code in one terminal, Copilot in their editor, and ChatGPT in a browser tab, the idea of consolidating those into one tool is appealing. The question is whether a single binary can do all of those things well enough to replace purpose-built alternatives. Install is a one-liner: curl -fsSL https://get.lukan.ai/install.sh | bash.