YV17labs/GhostDesk
MCP server that gives AI agents full control of a virtual Linux desktop — mouse, keyboard, screenshots, UI reading, clipboard, and shell. Human-like input simulation to bypass bot detection. Runs in Docker.
Platform-specific configuration:
{
"mcpServers": {
"GhostDesk": {
"command": "npx",
"args": [
"-y",
"GhostDesk"
]
}
}
}Add the config above to .claude/settings.json under the mcpServers key.
<p align="center"> </p>
<h1 align="center">GhostDesk</h1>
<p align="center"> <strong>Give your AI agent eyes, hands, and a full Linux desktop.</strong><br> An MCP server that lets LLM agents see the screen, move the mouse, type on the keyboard, read UI elements, fill forms, launch apps, and run shell commands — all inside a sandboxed virtual desktop. </p>
<p align="center"> <em>If a human can do it on a desktop, your agent can too.</em> </p>
<p align="center"> </p>
---
Most AI agents are trapped in text. They can call APIs and generate code, but they can't use software. GhostDesk changes that.
Connect any MCP-compatible LLM (Claude, GPT, Gemini...) and it gets a full Linux desktop with 25+ tools to interact with any application — browsers, IDEs, office suites, terminals, legacy software, internal tools. No API needed. No integration required. If it has a UI, your agent can use it.
Your agent gets its own Linux desktop. Here's what that unlocks:
"Go to the CRM, export last month's leads as CSV,
open LibreOffice Calc, build a pivot table,
take a screenshot of the chart, and email it to the teamLoading reviews...