Command IQ (CIQ) — Natural Language to Linux Commands

Command IQ (CIQ) is an offline, inference-based AI system that converts natural language instructions into Linux commands using pretrained FAISS and T5 models.

---

About AI Usage

Command IQ does not use online AI services (such as ChatGPT or OpenAI APIs).

All predictions are generated locally and offline using:

• Semantic vector search (FAISS)
• A pretrained transformer model (T5) for inference

No runtime training or external AI calls are performed.

⚡ Features

• Convert English instructions into working Linux commands
• Works fully offline using prebuilt FAISS + T5 model assets
• CPU-first design — GPU is optional, not required
• Supports Kali Linux, Ubuntu, Debian, WSL, and virtual machines
• Global CLI available from any directory using the ciq command
• Automatic setup using a single installer script (install_ciq.sh)
• Idempotent installer — safe to run multiple times

---

🚀 Quick Installation

1️⃣ Clone the repository

git clone https://github.com/ManojMJ17/Command_IQ.git
cd Command_IQ

---

2️⃣ Install Python 3.11 (required)

⚠️ Important Python 3.11 is required. Python 3.12 / 3.13 are not supported due to PyTorch and FAISS compatibility issues.

sudo apt update
sudo apt install -y \
  python3.11 \
  python3-dev \
  python3-pip \
  build-essential \
  curl unzip

---

3️⃣ Run the installer

chmod +x install_ciq.sh
./install_ciq.sh

The installer will:

• Create a Python 3.11 virtual environment
• Install all required dependencies (CPU-only PyTorch)
• Download & extract FAISS index and T5 model
• Create the global ciq CLI command

---

Optional Installer Flags

If assets are already downloaded:

./install_ciq.sh --no-download

Force re-download of assets:

./install_ciq.sh --force-download

---

🖥️ Verify Installation

ciq "check disk usage"

Example output:

Query         : check disk usage
Final Suggest : df -h

---

🧠 CPU vs GPU Support

Command IQ works fully on CPU.

• FAISS is used in CPU mode
• T5 model is used for inference only
• GPU / CUDA is optional, not required

CIQ works perfectly on:

• Kali Linux
• Virtual machines
• WSL
• Systems without NVIDIA GPUs

---

🐉 Kali Linux Notes (Important)

• Kali Linux ships with Python 3.13 by default
• CIQ requires Python 3.11
• The installer handles this automatically

If ciq is not found after installation (zsh users):

echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc
source ~/.zshrc

---

📦 Download Size Notice

• FAISS index: ~200 MB
• T5 model assets: ~1.1 GB

On slower or mobile networks:

• Downloads may take several minutes
• The installer supports resume on interruption

---

🧪 Usage

ciq "<natural language query>"

Examples

ciq "install VLC media player"
ciq "list all files recursively"
ciq "show current disk usage"

The CLI displays:

• FAISS suggestion
• T5 model suggestion
• Final merged Linux command
• Optional confirmation before execution

---

📂 Directory Structure After Installation

~/.ciq/
├─ src/
│  ├─ cli/
│  │  ├─ __init__.py
│  │  ├─ main.py
│  │  └─ predict.py
│  ├─ model/
│  │  ├─ saved_model/t5_base_resumed.pt
│  │  ├─ t5_base_arch/
│  │  └─ t5_base_tokenizer/
│  └─ faiss_index/
│     ├─ faiss_index_combined.index
│     └─ faiss_metadata_combined.pkl
├─ venv/
└─ bin/
   └─ ciq   (global CLI wrapper)

---

🧰 Troubleshooting

ciq: command not found

export PATH="$HOME/.local/bin:$PATH"

Python not found

sudo apt install python3.11 -y

Dependency installation failed

./install_ciq.sh

---

📜 License

MIT License — see the LICENSE file.

---

🧠 Author

Manoj Kumar GitHub: https://github.com/ManojMJ17

---