🔥 RAGE BORN — AI-Powered Terminal Agent

---

Author	Rohit Yadav
Institue	NIT Jalandhar
GITHUB	YADAV1825
HuggingFace	AutonomousX

---

RAGE is an AI-powered terminal agent for Linux. Think Claude Code or Gemini CLI, but designed to be your complete AI operating companion — automating terminal tasks, browser workflows, system administration, and more.

██████╗  █████╗  ██████╗ ███████╗    ██████╗  ██████╗ ██████╗ ███╗   ██╗
██╔══██╗██╔══██╗██╔════╝ ██╔════╝    ██╔══██╗██╔═══██╗██╔══██╗████╗  ██║
██████╔╝███████║██║  ███╗█████╗      ██████╔╝██║   ██║██████╔╝██╔██╗ ██║
██╔══██╗██╔══██║██║   ██║██╔══╝      ██╔══██╗██║   ██║██╔══██╗██║╚██╗██║
██║  ██║██║  ██║╚██████╔╝███████╗    ██████╔╝╚██████╔╝██║  ██║██║ ╚████║
╚═╝  ╚═╝╚═╝  ╚═╝ ╚═════╝ ╚══════╝    ╚═════╝  ╚═════╝ ╚═╝  ╚═╝╚═╝  ╚═══╝

✨ Features

• 🤖 Natural Language Interface — Just tell RAGE what you want in plain English
• 🛠 Tool System — 25 modular tools (terminal, files, git, browser, and Google Workspace)
• 📟 Multi-Terminal Tabs — Multiple named terminal sessions running in parallel
• 🌐 Browser Automation — Playwright-powered with persistent profile (logins, cookies retained!)
• 🏢 Google Workspace Integration — Automate Docs, Sheets, Gmail, Meet, and Calendar naturally
• 🔍 Web Search — Google search, page reading, link extraction, screenshots
• 🛡 Safety System — Detects dangerous commands and asks for confirmation
• ⏳ Background Tasks — Run long operations in the background with progress tracking
• 📝 Persistent Memory — Remembers your preferences and context across sessions
• 📜 Task History — Full log of everything RAGE has done
• 🎨 Premium UI — Beautiful terminal interface with colors, panels, and markdown rendering

🚀 Quick Install

One-liner

curl -fsSL https://raw.githubusercontent.com/YADAV1825/rage/main/install.sh | bash

pip

pip install rage-cli

From source

git clone https://github.com/YADAV1825/rage.git
cd rage
pip install -e .

⚡ Usage

Interactive Mode

rage

Direct Command

rage "Install Docker and configure it"
rage "Open example.com and tell me what it says"
rage "Search Google for Python tutorials"

Slash Commands

Command	Description
/help	Show help menu
/plan	Refines the prompt make a task.md and follows it to finish task better
/tasks	Show background tasks
/history	Show command history
/memory	Show what RAGE remembers
/terminals	Show active terminal sessions
/browser	Show browser tabs & status
/browseurl	Open a URL in the browser
/searchq	Google search from terminal
/glogin	Log in to Google Workspace
/model	Select AI model (LLM)
/config	Show configuration
/clear	Clear screen
/exit	Exit RAGE BORN

Browser Capabilities

RAGE can automate a full Chromium browser with a persistent profile:

# Ask RAGE BORN naturally
❯ Open google.com and search for "best Linux distros 2026"
❯ Go to github.com and check my notifications
❯ Fill out the form on that page with my details
❯ Take a screenshot of this page
❯ Read the article on that website and summarize it

Persistent profile means you only log in once — cookies, sessions, and bookmarks are saved in ~/.rage/browser_profile/. Run /glogin to sign in to Google.

🔑 Configuration

On first run, RAGE BORN will ask for your Lightning AI API key.

You can also set it via environment variable:

export RAGE_API_KEY="your-key-here"

Configuration is stored in ~/.rage/config.json.

🏗 Architecture

User → Planner (LLM) → Task Manager → Tool Manager → Tools (25 total)
                                                        ├── Terminal (run_command, list_terminals)
                                                        ├── File System (read, write, list, search)
                                                        ├── Git (status, commit, push, etc.)
                                                        ├── Browser (13 generic web tools)
                                                        ├── Google Workspace (5 macro tools)
                                                        │   ├── gworkspace_docs, gworkspace_sheets
                                                        │   └── gworkspace_gmail, gworkspace_meet, calendar
                                                        └── Future Plugins

🛡 Safety

RAGE detects dangerous commands and requires explicit approval:

• rm -rf / recursive deletion
• dd / disk operations
• mkfs / filesystem formatting
• shutdown / reboot
• Firewall changes
• Piping remote scripts to shell

You can approve once, or add patterns to an "always allow" list.

📂 Project Structure

rage/
├── __init__.py          # Package info
├── __main__.py          # python -m rage entry
├── cli.py               # Interactive REPL
├── agent.py             # Core AI agent loop
├── llm.py               # LLM API client
├── config.py            # Configuration
├── memory.py            # Persistent memory
├── history.py           # Command history
├── safety.py            # Dangerous command detection
├── banner.py            # ASCII art banner
├── ui.py                # Rich terminal UI
├── tools/
│   ├── base.py          # Base tool class
│   ├── registry.py      # Tool registry (20 tools)
│   ├── terminal.py      # Shell execution + multi-tab
│   ├── filesystem.py    # File operations
│   ├── git_tool.py      # Git integration
│   └── browser.py       # Playwright browser (13 tools, persistent profile)
└── tasks/
    ├── manager.py       # Background task manager
    └── models.py        # Task data models

📄 License

MIT License — see LICENSE for details.

---

Built with 🔥 RAGE BORN