Terminal Basics for AI Developers — the 10 Commands You Actually Need

10 commands. That's all that stands between you and your first local LLM.

No CS degree, no Linux expertise required. Anyone who wants to run Ollama, check Docker containers, and download models needs exactly these 10 commands — nothing more.

Why the Terminal at All?

Because almost every AI tool runs from the command line. Docker runs there. Python scripts start there. SSH connections to your server (when you eventually move to dedicated hardware) go through the terminal. Avoiding it means spending half your time wrestling with GUI wrappers that just call the terminal anyway. Direct is faster.

On Windows we recommend Windows Terminal + WSL2 — that gives you a full Linux environment without dual-booting.

The 10 Commands

Commands in Practice — Real Examples

Each command below shows at least two real-world use cases you will encounter when running a local AI stack.

cd — navigate to your project:

cd ~/projects/ai-stack          # jump to your AI project
cd ..                           # go up one level
cd -                            # jump back to previous directory

ls — see what's in a directory:

ls -la                          # show all files including hidden, with details
ls -lh ~/.ollama/models/        # show model files with human-readable sizes
ls *.yml                        # list only YAML files (e.g., docker-compose)

curl — test APIs directly from the terminal:

curl http://localhost:11434/api/tags                    # list Ollama models
curl -X POST http://localhost:11434/api/generate \
  -d '{"model":"qwen3.5:4b","prompt":"Hello"}'         # send a prompt
curl -s http://localhost:3000/health | python -m json.tool  # check Open WebUI health

grep — find what matters in noisy output:

docker logs ollama | grep "error"              # find errors in Ollama logs
grep -r "OLLAMA_HOST" ~/projects/              # find config references in your project
docker logs open-webui | grep -i "warning"     # case-insensitive warning search

docker ps — know what's running:

docker ps                                              # running containers
docker ps -a                                           # all containers including stopped
docker ps --format "table {{.Names}}\t{{.Status}}\t{{.Ports}}"  # clean overview

Three Things That Help Immediately

Use Tab completion. Start typing cd pro and press Tab — the terminal completes projects/ automatically. Saves typing and prevents typos.

Arrow key up. The last command comes back with ↑. No retyping needed. Press Ctrl+R to search through your command history.

| (pipe) connects commands. docker logs ollama | grep error shows only lines containing "error" — instead of scrolling through hundreds of log lines. You can chain multiple pipes: docker logs ollama | grep error | tail -5 shows only the last 5 error lines.

Windows Users: WSL2 Is Worth It

Ollama runs natively on Windows. But many tools in the AI ecosystem are built for Linux. WSL2 gives you a full Ubuntu environment directly inside Windows — no VM, no dual-boot.

# Install WSL2 (PowerShell as Admin)
wsl --install

After a restart you have Ubuntu available in Windows Terminal. All 10 commands above work identically. Details: WSL2 Documentation.

For a deeper dive: the Linux Foundation has free introductory materials for the command line.

Quick Test: Is Your Terminal Working?

Open Terminal (Windows Terminal, macOS Terminal, or a Linux shell) and type:

curl --version

If a version number appears — perfect. If not, curl isn't installed. On Windows: winget install curl.curl. On Ubuntu/WSL2: sudo apt install curl.

Terminal working? Then comes the real step: install Ollama and run your first model locally.

Continue to Step 3: Install Ollama in 5 Minutes →

Or jump straight to the complete setup guide: the Local AI Playbook P1 (EUR 49) takes you from terminal to a production-ready stack with browser interface, API access, and pre-configured Docker containers.