🎤 AI RUN CMD

A fun and functional terminal AI assistant using ChatGPT, Claude, or local models via Ollama. Just type ai do something and let DMC help you drop command-line hits. Works with cloud-based and local AI models on Linux, macOS, and Windows (WSL).

---

🚀 Features

See features.md for a complete list of features and capabilities.

---

🛠 Installation

Linux / macOS

bash <(curl -s https://raw.githubusercontent.com/kenshub/ai-run-cmd/main/scripts/install.sh)

Clones the repo, installs dependencies, sets up .env, and updates your .bashrc or .zshrc. On macOS, Homebrew will be installed automatically if not already present.

Windows (WSL)

Download and run install.ps1 in PowerShell:

irm https://raw.githubusercontent.com/kenshub/ai-run-cmd/main/install.ps1 | iex

Installs WSL with Ubuntu if needed (requires a reboot), then runs the standard installer inside WSL.

---

See install.md for manual setup instructions.

---

🤖 AI Providers

Provider	Requires
OpenAI (default)	API key
Anthropic (Claude)	API key
Groq	API key
Mistral	API key
Google Gemini	API key
Ollama (local)	No key — runs on your machine

Switch providers anytime:

ai provider ollama
ai provider openai

---

💻 Local LLM (No API Key)

Run AI entirely on your machine — no internet or API key required:

ai install-local

This installs Ollama and lets you choose a model:

Model	Size	Best for
qwen2.5:0.5b	~400MB	Fastest, everyday commands
llama3.2:1b	~1.3GB	Good balance
phi3:mini	~2.2GB	More capable responses

---

🧪 Usage

ai restart apache
ai run docker prune
ai list files by size

Special commands:

ai explain tar -czpf         # explain what a command does
ai rap hard drive space       # get the answer in rap format
ai install-local              # set up a local LLM
ai provider ollama            # switch AI provider
ai debug on                   # enable debug output

---

☕ Buy Me a Coffee

If this helped you out:

Buy me a coffee

Or just ⭐ the repo and tell a friend.