Arawn is designed for edge computing — a single binary with embedded storage.
The quickest way to install Arawn and its dependencies:
curl -fsSL https://raw.githubusercontent.com/colliery-io/arawn/main/scripts/install.sh | shThis will:
- Download the latest release binary from GitHub
- Install sandbox dependencies on Linux (bubblewrap, socat)
- macOS sandbox support is built-in — no extra dependencies needed
Options:
# Install a specific version
./install.sh --version v0.1.0
# Install to a custom directory
./install.sh --install-dir /usr/local/bin
# Skip sandbox dependency installation
./install.sh --skip-deps
# Preview what would happen
./install.sh --dry-run- Rust 1.85+ with cargo
- C compiler (for SQLite bindings)
- Linux only: bubblewrap and socat for sandboxed shell execution
- Optional: ONNX Runtime for local embeddings
# Clone the repository
git clone https://github.com/colliery-io/arawn.git
cd arawn
# Build release binary
cargo build --release
# Binary location
./target/release/arawnEnable ONNX features for local embedding generation:
# Build with GLiNER for local NER + embeddings
cargo build --release --features gliner
# Models downloaded on first use:
# - all-MiniLM-L6-v2 (embeddings, ~80MB)
# - GLiNER-small (NER, ~400MB)Arawn requires bubblewrap and socat on Linux for sandboxed shell execution:
# Ubuntu/Debian
sudo apt-get install bubblewrap socat
# Fedora
sudo dnf install bubblewrap socat
# Arch
sudo pacman -S bubblewrap socat
# Alpine
sudo apk add bubblewrap socat
# openSUSE
sudo zypper install bubblewrap socatArawn uses the following data layout:
~/.arawn/
├── arawn.toml # User configuration
├── arawn.log # Rotating logs
└── data/
├── memory.db # SQLite (memories, sessions, notes)
├── memory.db-wal # Write-ahead log
├── workstreams.db # Workstream cache
└── workstreams/ # JSONL message history
| Resource | Minimum | Recommended |
|---|---|---|
| RAM | 512 MB | 2 GB (with local embeddings) |
| Disk | 100 MB | 1 GB (for memories + models) |
| CPU | 2 cores | 4 cores (parallel tool execution) |
# Check version
arawn --version
# Verify configuration
arawn config show
# Test with a simple question
arawn ask "Hello, are you working?"- Quick Start — Your first conversation
- Configuration — Set up LLM backends