agent-desktop
by Vercel
Official listing: built and published by Vercel. More from Vercel
Control macOS, Windows and Linux desktop apps from AI agents through the OS accessibility tree, with a JSON CLI, MCP server and C library for computer-use builders.
Use it now
Keep it with up to 9 other tools you're setting up, then share them all as one link.
Try this first
Ask your assistant, with agent-desktop connected:
- Accessibility Snapshot
Use agent-desktop to read a desktop app's accessibility tree: ‹what you need›
- Act on Elements
Use agent-desktop to click UI elements: ‹what you need›
- Apps and Windows
Use agent-desktop to launch and clos apps: ‹what you need›
Connect it over MCP
Connect agent-desktop as an MCP server, following its setup instructions at https://github.com/vercel-labs/agent-desktop#quick-start
Setup guide: github.com/vercel-labs/agent-desktop
Install the CLI
npm install -g agent-desktop
Setup guide: github.com/vercel-labs/agent-desktop
Compare
Press Compare to add it to the tray at the bottom of the window, then pick up to 2 others anywhere on the site.
- Access
- CLIMCP
- Auth
- None
- Skills
- 5
Skills
Accessibility Snapshot
Reads a desktop app's accessibility tree and returns it as JSON, with a ref for every element the agent can act on.
Act on Elements
Clicks, types, selects, toggles and scrolls UI elements by ref, headless by default without moving the pointer.
Apps and Windows
Launches and closes apps, lists windows and displays, and focuses, resizes, moves, minimizes or maximizes windows.
Clipboard and Notifications
Reads, sets and clears the clipboard, and lists, dismisses or triggers actions on desktop notifications.
MCP Server
Serves every agent-desktop command as an MCP tool over stdio and returns each command's JSON envelope.
