# agent-desktop (Vercel Labs' CLI for AI agents to operate desktop apps)

> Updated 2026-11-02 · type: tool · category: mcp-agents · status: active · rev 1

agent-desktop lets an AI agent click through native Mac, Windows or Linux apps by reading their accessibility tree, not screenshots.

- Open source: yes (Apache-2.0)
- Self-hostable: yes
- Pricing: Free (Apache-2.0), local binary, no paid tier. Optional helper sends screen descriptions to a model provider you configure.
- Best for: Teams automating work in desktop-only apps (legacy CRMs, finance tools, native clients) with an AI agent, who prefer structured element refs over screenshot clicking.
- Not for: Teams whose systems all have APIs or web UIs (use the API or a browser agent), and anyone uneasy giving an agent control of a logged-in desktop.
- Version: 1.0.2
- Last verified: 2026-11-02

- **Canonical:** https://gtmstacker.com/registry/tool/agent-desktop/
- **Source:** [Introducing agent-desktop (agent-browser, for your whole computer)](https://x.com/ctatedev/status/2116323148706771302) — ctatedev
- **Tags:** mcp-agents, computer-use, desktop-automation, accessibility-tree, rust, vercel, open-source
- **Repository:** https://github.com/vercel-labs/agent-desktop
- **Version:** 1.0.2

## Is agent-desktop (Vercel Labs' CLI for AI agents to operate desktop apps) open source?

Yes, agent-desktop (Vercel Labs' CLI for AI agents to operate desktop apps) is open source under the Apache-2.0 license.

## How much does agent-desktop (Vercel Labs' CLI for AI agents to operate desktop apps) cost?

Free (Apache-2.0), local binary, no paid tier. Optional helper sends screen descriptions to a model provider you configure. (free).

## Can I self-host agent-desktop (Vercel Labs' CLI for AI agents to operate desktop apps)?

Yes, agent-desktop (Vercel Labs' CLI for AI agents to operate desktop apps) can be self-hosted (the source is available under the Apache-2.0 license).

## Alternatives & related

- [cua (Computer-Use Agents)](https://gtmstacker.com/registry/tool/cua-computer-use-agents/)
- [browser-use](https://gtmstacker.com/registry/tool/browser-use/)


---

**agent-desktop** (`vercel-labs/agent-desktop`) is an **Apache-2.0** Rust CLI (with a C FFI) that lets AI agents operate native desktop applications on macOS, Windows and Linux. It reads each app's **OS accessibility tree** and returns **JSON snapshots with element refs** the agent can act on, so it does not depend on screenshots and pixel coordinates. It is headless by default and exposes 56 operational commands. It comes from the same Vercel Labs team as agent-browser.

## Provenance

- License **Apache-2.0** (GitHub license API on the LICENSE file; npm license field agrees).
- Identity: `vercel-labs` org, same org as `vercel-labs/agent-browser` (45,001 stars); homepage agent-browser.dev/desktop.
- Activity as of 2026-11-02 (GitHub API): **8,774 stars, 610 forks**, 120 open issues, last push 2026-11-02 (up from 3,680 stars on 2026-10-31); releases v1.0.0 and v1.0.1 (2026-10-28), v1.0.2 (2026-10-30). npm `agent-desktop` 1.0.2: 2,941 downloads 2026-09-30 to 2026-10-29.
- **Vendor claim:** "stable refs / safe to retry" is from the README.
- Surfaced via the author's X post in the 2026-10-31 pull (the post returned HTTP 402; the repo was resolved through the GitHub API). Curated and verified in run `2026-10-31-registry-update-01`; re-verified on the GitHub API and published in run `2026-11-03-registry-update-01`.

## Why it matters for a GTM stack

Some revenue work still happens in desktop apps with no API: an old on-prem CRM, a finance client, a dialer. agent-desktop gives an agent a structured way to work those apps. Start with read-only tasks such as copying records into a sheet, and keep it away from anything that sends or deletes until you have watched it run.