Vercel Labs Open Sources Agent‑Desktop: Token‑Efficient OS AI Agent
Vercel Labs releases the source code of agent‑desktop, an OS‑level AI agent that reads the accessibility tree, uses few tokens, runs in background, and supports multiple cursors.
Vercel Labs has opened the source code of agent-desktop.
The tool brings the Computer Use approach, familiar from agent-browser, to the entire operating system, and is characterized by economical token usage.
The main feature of the tool is that it does not require screenshots. Instead of analyzing images, the agent directly reads the system's accessibility tree, obtaining information about buttons, input fields, and other interface elements.
At the same time, agent-desktop runs in the background and does not intercept mouse control, so the user can continue working on the computer. Simultaneous operation of multiple agents is also supported, each receiving its own cursor.
Grok Bot is happy.
Comments
Weekly digest
The best of vibe coding, AI agents and open source — once a week, no spam.
Related articles
All articlesCline Desktop Launches AI Agent Workspace for macOS & Windows
Cline Desktop offers an open‑source AI workspace with parallel agents, task scheduling, checkpoints, branching, and a plugin marketplace, now on macOS and Windows.
Tolaria: Free macOS Markdown Knowledge Base App Built with AI Agents
Tolaria is a free, open-source macOS desktop app for Markdown knowledge bases and collaboration—built in 3 months as a major AI-assisted coding experiment.
Google Opens Harness²: Open‑Source Framework for Recursive AI Agents
Google releases Harness², an open‑source harness enabling recursive management and learning for AI agents with system prompts, memory, tools, subagents, and plugins.