Installation¶
Prerequisites¶
| Requirement | Version | Notes |
|---|---|---|
| OS | macOS 12+ or modern Linux | Windows is not supported yet |
| Python | 3.9+ | python3 --version to check |
| Ollama | Latest | Local LLM runtime |
| Shell | Any | ZSH is only needed for optional Ctrl+Space hotkey integration |
| Git | Recent | Recommended for update workflow and project-aware context |
Install scope
Normal install (pip install -e .) only installs runtime dependencies.
It does not install test or docs tooling.
Install Ollama¶
Drift uses Ollama for local LLM inference.
Auto-Setup (default)
You don't need to install Ollama manually. When you run any drift command for the first time, it will automatically:
1. Install Ollama (via Homebrew on macOS, or the official script on Linux)
2. Start the Ollama server
3. Pull the default model (`qwen2.5-coder:1.5b`)
To disable this, set `auto_install_ollama`, `auto_start_ollama`, or `auto_pull_model` to `false` in `~/.drift/config.json`.
If you prefer to install manually:
# macOS — download from ollama.com or:
brew install ollama
# Start Ollama
open -a Ollama # or: ollama serve
Pull the default model:
Model choice
qwen2.5-coder:1.5b (~1GB) is fast and works well for command generation. For better quality, try qwen2.5-coder:7b (~4.5GB) or codellama:7b.
Install Drift CLI¶
Verify Installation¶
Expected output:
✓ Ollama is installed
✓ Ollama is running
✓ Model qwen2.5-coder:1.5b is available
auto_install_ollama: ON
auto_start_ollama: ON
auto_pull_model: ON
✓ Drift directory exists (~/.drift)
ZSH Integration (Optional)¶
The installer sets up a Ctrl+Space hotkey so you can type natural language directly in your terminal:
Then restart your terminal. Type a natural language query and press Ctrl+Space.
Next Steps¶
- Quick Start → — Try your first commands
- Configuration → — Customize Drift settings