tuxgenie 3.9.0

AI-powered Linux assistant — fix any Linux problem in plain English

AI Terminal

An intelligent Linux troubleshooter that accepts natural language problem descriptions and fixes them automatically using the Anthropic Claude API.

What it does

Describe your Linux problem in plain English. AI Terminal will:

1. Collect system context (OS, kernel, user, desktop, uptime) silently at startup
2. Send the issue + system info to Claude via the Anthropic API
3. Receive a structured fix plan with step-by-step commands
4. Show each step with its risk level before running it
5. Ask for confirmation before executing anything
6. Feed command outputs back to Claude and iterate until resolved (up to 6 rounds)

Quick Start

1. Install

git clone <repo>
cd ai-terminal
bash install.sh

2. Set your API key

export ANTHROPIC_API_KEY="sk-ant-..."

Get your key at console.anthropic.com.

3. Run

aifix

Or without installing:

ANTHROPIC_API_KEY=sk-ant-... python3 ai_terminal.py

Example issues you can describe

• nginx won't start after I changed the config
• my disk is almost full, help me find what's taking space
• SSH connections keep timing out
• a process is eating 100% CPU, I don't know which one
• my cron job isn't running
• docker containers can't reach the internet
• I accidentally changed file permissions, nothing works

Safety

AI Terminal never runs a command without your explicit approval. For each step you see:

• Risk level — SAFE, MODERATE, or DANGEROUS (colour-coded)
• [SUDO NEEDED] badge when root is required
• A red warning banner for destructive commands
• A prompt: [y/n/skip/abort] — you decide what runs

Commands matching dangerous patterns (rm -rf /, dd if=, mkfs, fdisk, wipefs, shred, chmod 777 /, fork bombs) are flagged regardless of what Claude labels them.

How it works

User describes issue
       │
       ▼
  System context collected (OS, kernel, user, uptime…)
       │
       ▼
  Claude API → structured JSON fix plan
       │
       ▼
  Display steps with risk badges
       │
       ▼
  User approves/skips each command
       │
       ▼
  Command output captured and sent back to Claude
       │
       ▼
  Claude iterates (max 6 rounds) until resolved

File listing

ai-terminal/
├── ai_terminal.py      # Main script — run this directly
├── install.sh          # Installer (creates ~/.local/bin/aifix)
├── requirements.txt    # Python dependencies
└── README.md           # This file

Requirements

• Python 3.8+
• anthropic Python package (auto-installed by the script)
• An Anthropic API key
• Linux (the tool is designed for Linux environments)