What is OpenAgents Launcher?
OpenAgents Launcher is a desktop app that installs, configures, and runs AI agents on your computer — no terminal required.
What is OpenAgents Launcher?
The OpenAgents Launcher is a desktop app (macOS, Windows, Linux) that manages AI agents on your computer. Think of it as the app store plus control panel for coding agents:
- Install agents with one click — Claude Code, Codex, Cursor, Aider, Goose, Gemini CLI, and more from a built-in catalog. The Launcher handles the messy parts (dependencies, versions — it will even install Node.js for you if it's missing).
- Configure everything in a GUI — API keys, models, and endpoints in a settings form instead of environment variables and config files.
- Run and monitor — see every agent in one list with a live status (running / stopped), start and stop them with a click, and read their logs.
- Connect to a Workspace — link your local agents into an OpenAgents Workspace so you (and your team) can talk to them from the browser.
If you've ever spent an afternoon fighting npm, pip, and API-key environment variables just to try a coding agent — the Launcher is the "it just works" alternative.
You don't need a workspace to use the Launcher — it's a complete local agent manager on its own. Connecting a Workspace adds the team layer: reach your agents from any browser and collaborate with others. (Workspace vs Launcher →)
Who is it for?
- Anyone trying agents for the first time. Download, click install, paste an API key, done.
- People who run agents daily and want them managed in one place — updated, configured, and running in the background.
- Teams using the Workspace — the Launcher is the easiest way to get your local machine's agents into a shared workspace.
Finding your way around
The sidebar walks you through the flow in order — it even labels the steps:
| Section | What it's for |
|---|---|
| Dashboard | The overview: what's running, messages today, fleet health, recent activity, quick actions. |
| Marketplace | Step 1 — browse and install agents (categories, search, one-click install with live progress). |
| Agents | Step 2 — your installed agent instances: create, configure keys, test, start/stop, join workspaces. |
| Workspaces | Step 3 — the workspaces this machine serves: health, agents, open in browser. |
| Chat | Talk to your workspace agents right from the desktop app. |
| Connections & Credentials | Connect platforms (GitHub, Slack, Notion, …) and store API keys once — encrypted on this machine — then apply them to any agent. |
| Logs | Live activity and errors across all agents, with filters and a timeline view. |
| Settings | Appearance, notifications, network (including a China mirror), updates, and more. |
Press ⌘K / Ctrl+K anywhere for the command palette — jump to any page, start/stop agents, or switch theme.
The engine underneath
Behind the app, a small background service (the daemon) keeps your agents running — even with the window closed. That's what lets agents stay online in a workspace 24/7. Power users can drive the same engine from the terminal with the agn CLI (CLI Reference) — but you never need to.
Free and open source
Like the rest of OpenAgents, the Launcher is open source and free. You only pay your AI provider for the model calls your agents make.
Next
Install the Launcher on your machine.