[Open Source] xAgent CLI - The FIRST AI assistant that can actually control your desktop
I've been building something unique - an AI CLI tool that doesn't just read/write files, but can actually control your mouse and keyboard. It's called xAgent CLI.
Why is this a big deal?
Most AI coding tools (Claude Code, Cursor, Copilot) can only:
read_file and write files
Execute shell commands
Search codebases
But they can't:
Click buttons on screen
Fill out web forms
Navigate websites
Control desktop apps
xAgent CLI can do ALL of this.
Key Features:
🖱️ True GUI Automation
Precise mouse coordinate control
Keyboard input simulation
Browser automation
Control ANY application on your PC
🧠 Access to Frontier Models
MiniMax M2.1 (High-performance reasoning and coding)
GLM-4.7 (From Zhipu AI)
Kimi K2 (MoE model from Moonshot AI)
Qwen3 Coder (Alibaba's coding model)
ALL FREE, no API keys needed
💻 Developer Tools
Code analysis and refactoring
Bug detection and fixing
Project architecture analysis
Context compression for large repos
🏠 Life Automation
"Organize my desktop"
"Download all PDFs from this page"
"Set up daily backups"
"Find and remove duplicate files"
🔒 Security Modes
5 modes from YOLO (full access) to DEFAULT (approval required)
Example Usage:
xagent gui --url https://example.com
> Click the login button at coordinates (500, 300)
> Type "myemail@example.com" in the username field
> Type "mypassword" in the password field
> Click the submit button
Quick Start:
npm i -g @xagent-ai/cli
xagent start
Cross-platform: Windows, macOS, Linux.
Would love your feedback!
Replies