Documentation
Localix Terminal docs
How to install Localix Terminal, connect your coding agents, and work in the window: terminal sessions, the browser, the board, the Store, and the one command agents use.
Start
From nothing to an agent working in a project.
- Getting startedInstall the app, open a folder as a project, and open a terminal or start an agent in it. The app is all there is to install: it holds the window, the llxterm command, the MCP server and, on macOS…
- Connecting agentsThe app connects the coding agents you have installed by itself, and keeps them connected. It has no model and no account of its own: the agents are the ones you already run.
In the window
What a project has open, and how each part behaves.
- Terminals and sessionsA terminal is a real shell on a PTY, and a coding agent running in one is a session. Agents stay the command-line programs their makers ship, with their own commands, sign-in and updates.
- The browserPages open as tabs on the system WebView: the project's own pages, a dev server, or any site. You and an agent look at the same page.
- The boardThe project's kanban: Backlog, Todo, Doing, Review, Done and Discarded. You hand cards to agents, and only you accept what they hand in.
- The StorePlugins, skills and MCP servers for your coding agents in one searchable list (⇧⌘S). Nothing is installed before you are shown what it would write and run.
- Files and the editorThe project's tree is in the side, under its tabs, live as an agent or you change it. A file picked in it opens in an editor tab.
- Projects and windowsA project is a folder. Its sessions, tabs, history, board and events are its own, and it opens as you left it, even after a restart.
- Shortcuts and settingsEvery command of the window has a key, and every key is yours to change. Settings are one tab, opened with ⌘,.
For agents
The one command an agent or a script works in the window with.
Reference
The switches, the protocol, and what the app does and does not allow.
- ConfigurationThe app needs no configuration to run. These are the few switches it has, for secrets, for where it keeps its files, and for running with no window.
- The app and its protocolThe app is the server: it owns the window, the terminals, the board and the database, and serves one API on a unix socket. Everything else is a thin face of that API with nothing of its own.
- SecurityThe app has no approval prompt of its own: the agent's permission to run the command is the approval. Allowing llxterm window * lets an agent open pages, run scripts in them and start programs as…