Skip to main content

Installation

Not Yet Public

OWL is currently in development and not open for public installation. This documentation is for presentation and demonstration purposes.

Get OWL running with a single command.

curl -sSL https://raw.githubusercontent.com/kavinps/owl/master/install.sh | bash

This script automatically:

  • Installs Ollama if not present
  • Pulls required models (llama3.2 and nomic-embed-text)
  • Installs uv package manager
  • Installs OWL globally

After installation, run from any directory:

owl

Manual Installation​

If you prefer manual setup:

1. Install Ollama​

# Linux
curl -fsSL https://ollama.ai/install.sh | sh

# macOS
brew install ollama

2. Pull Models​

# Start Ollama
ollama serve

# Pull models (in another terminal)
ollama pull llama3.2           # Chat model
ollama pull nomic-embed-text   # Embeddings for knowledge base

3. Install OWL​

Using uv (recommended):

# Install uv
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install OWL globally
uv tool install git+https://github.com/kavinps/owl.git

Or using pip:

pip install git+https://github.com/kavinps/owl.git

Verify Installation​

owl

You should see:

  (o,o)
  /)_)   OWL - A wise AI with a growing soul
  " "

Project: your-project-name
Type /help for commands, or just chat!

Commands​

owl              # Start interactive session (auto-starts daemon)
owl stop         # Stop daemon
owl status       # Check daemon status

Data Directory​

OWL stores data in ~/.owl/:

~/.owl/
├── config.yaml      # Configuration
├── soul.yaml        # OWL's evolving character
├── memory.md        # Human-readable memory
├── memory/
│   └── owl.db       # SQLite database
├── knowledge/
│   └── chroma/      # Vector database
└── owl.sock         # Unix socket (when running)

Configuration​

Create ~/.owl/config.yaml to customize:

llm:
  provider: ollama
  host: http://localhost:11434
  model: llama3.2        # Or any Ollama model
  timeout: 180
  temperature: 0.7

daemon:
  log_level: INFO

Next Steps​