What this skill does
Desktop automation via native OS accessibility trees using the agent-desktop CLI. Use when an AI agent needs to observe, interact with, or automate desktop applications (click buttons, fill forms, navigate menus, read UI state, toggle checkboxes, scroll, drag, type text, take screenshots, manage windows, use clipboard, manage notifications). Covers 58 command names (54 operational; four held-input names fail closed until daemon ownership exists) across observation, interaction, keyboard/mouse, app lifecycle, notifications (macOS), clipboard, wait, session lifecycle, and a `skills` command that bundles docs straight from the binary. Triggers on: "click button", "fill form", "open app", "read UI", "automate desktop", "accessibility tree", "snapshot app", "type into field", "navigate menu", "toggle checkbox", "take screenshot", "desktop automation", "agent-desktop", or any desktop GUI interaction task. Supports the macOS Phase 1 adapter, with Windows and Linux planned against the same core contracts.
agent-desktop is part of the Research & Knowledge category — research and knowledge skills that gather, search, and summarize information. You can install it on its own or alongside other research & knowledge skills from the OpenClaw catalog.
How to install agent-desktop
The easiest path is via the OpenClaw Easy desktop app — one click, no terminal required:
- Download OpenClaw Easy for macOS or Windows (free, one-click installer, ~30 seconds).
- Open the in-app Skills panel.
- Search for
agent-desktopand click Install. - The skill activates automatically when an incoming message matches its description.
Install from the command line
If you already run the OpenClaw CLI, add agent-desktop with a single command:
openclaw skills add agent-desktop
This pulls agent-desktop from ClawHub and installs it into ~/.openclaw/skills/agent-desktop/. Restart the OpenClaw gateway afterwards so the new skill is discovered.
How to use agent-desktop
Once installed, agent-desktop activates on its own: when an incoming message on WhatsApp, Telegram, Slack, Discord, Feishu or Line matches the skill's description, your OpenClaw agent loads it and runs the workflow. You can also trigger it explicitly by describing the task in chat. No extra configuration is required after install.
Turning images into video
agent-desktop works with images. Getting from there to something postable is a separate job: ViralMint — an open-source video pipeline — turns stills into motion — pans and transitions, synced captions, and a vertical export for short-form feeds.
It runs as an MCP server, so an OpenClaw agent can drive it from the same chat you already use: ask for a short, and the render comes back finished. See how to connect a video pipeline to your agent.
Manual install (advanced)
If you prefer manual installation:
- Click the Download .zip button above to grab
agent-desktop-0.1.22.zipdirectly from our S3 mirror. - Unzip into
~/.openclaw/skills/agent-desktop/(create the directory if it does not exist). - Restart OpenClaw Easy (or the OpenClaw CLI gateway) so the new skill is discovered.
Frequently asked questions
How do I install agent-desktop?
Install agent-desktop in the OpenClaw Easy desktop app by opening the Skills panel, searching for agent-desktop, and clicking Install. From a terminal you can run: openclaw skills add agent-desktop. Either way the skill is placed in ~/.openclaw/skills/agent-desktop/.
Is agent-desktop free?
Yes. agent-desktop is free and open-source, distributed under the Apache-2.0 license through ClawHub. No account or payment is required to download or run it.
What does agent-desktop do?
Desktop automation via native OS accessibility trees using the agent-desktop CLI.
Related: more research & knowledge skills
If agent-desktop looks useful, you may also want to check out other research & knowledge skills in the OpenClaw catalog:
Browse the full OpenClaw skill catalog
This page covers just one skill. The OpenClaw skill hub has 10,000+ more — search, sort by downloads or stars, and install any of them in one click. There is also a curated awesome-openclaw-skills list grouped by use case.
Get OpenClaw Easy — Free
Install agent-desktop and 10,000+ other OpenClaw skills in one click. Free, open-source, runs locally on macOS & Windows.
Free, open-source · Apache-2.0 · Works with Claude, ChatGPT, Gemini, or local Ollama models