Skip to main content
Get OneClaw up and running in minutes. This guide will take you from building the binary to running your first AI agent.

​
Prerequisites

1

Install Rust

OneClaw requires Rust 1.85 or later with edition 2024 support.
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
rustup update
Verify your installation:
rustc --version
# Should show: rustc 1.85.0 or later
2

Clone the repository

git clone https://github.com/nclamvn/oneclaw.git
cd oneclaw

​
Build and Run

1

Build the binary

Build OneClaw in release mode for optimal performance:
cargo build --release
This produces a ~3.4MB binary in target/release/oneclaw-core.
2

Run OneClaw

Start the interactive CLI:
cargo run --release -p oneclaw-core
You’ll see the OneClaw prompt:
oneclaw>
3

Try basic commands

Test the system with these commands:
# View system status
oneclaw> status

# Store a memory
oneclaw> remember The project started in February 2026

# Search memories
oneclaw> recall project start

# View available commands
oneclaw> help

​
Configuration

1

Create configuration file

Create config/default.toml in your working directory:
mkdir -p config
2

Configure an LLM provider

Choose your provider and add configuration. OneClaw supports 6 providers:
[security]
deny_by_default = true

[provider]
primary = "anthropic"
model = "claude-sonnet-4-20250514"
max_tokens = 1024
temperature = 0.3
fallback = ["ollama"]

[provider.keys]
# Or set ANTHROPIC_API_KEY environment variable
# anthropic = "sk-ant-..."
3

Set API keys (if using cloud providers)

Set environment variables for your chosen provider:
# Anthropic
export ANTHROPIC_API_KEY="sk-ant-..."

# OpenAI
export OPENAI_API_KEY="sk-..."

# Google (Gemini)
export GOOGLE_API_KEY="AIza..."

# Or use the generic key
export ONECLAW_API_KEY="your-key-here"

​
First AI Interaction

With a provider configured, you can now use the ask command: 
oneclaw> ask What is OneClaw?

# OneClaw will route your question to the configured LLM provider
# and return a response based on its training data

​
Example Session

Here’s a complete example session showing OneClaw’s key features: 
# Start OneClaw
$ cargo run --release -p oneclaw-core

oneclaw> status
Runtime Status:
  Security: DefaultSecurity (deny-by-default)
  Memory: SqliteMemory (0 entries)
  Provider: Anthropic (claude-sonnet-4-20250514)
  Uptime: 0.12s

oneclaw> remember OneClaw is a 6-layer AI agent kernel for edge devices

oneclaw> recall agent kernel
Found 1 result:
  [2026-03-02 07:30] OneClaw is a 6-layer AI agent kernel for edge devices

oneclaw> ask What are the benefits of edge AI?
[LLM Response: Benefits of edge AI include low latency, data privacy, 
offline operation, reduced bandwidth costs, and real-time processing...]

oneclaw> providers
Available LLM Providers:
  ✓ anthropic (claude-sonnet-4-20250514) - Active
  ✓ ollama (llama3.2:3b) - Fallback
  ✗ openai - No API key
  ✗ deepseek - No API key

oneclaw> metrics
Operational Metrics:
  Messages processed: 4
  LLM calls: 1
  Memory operations: 2
  Average LLM latency: 247ms

oneclaw> exit
Shutting down...

​
Command Reference

CommandDescription
statusSystem overview with metrics
healthLayer-by-layer health check
remember <text>Store in memory
recall <query>Search memory (hybrid FTS5 + vector)
ask <question>Query LLM with context
providersList LLM provider status
toolsList available tools
channelsList active channels
helpShow all commands
exitGraceful shutdown
To enable semantic memory search with embeddings:
1

Configure embedding provider

Add to your config/default.toml:
[embedding]
provider = "ollama"
model = "nomic-embed-text"
2

Install embedding model (if using Ollama)

ollama pull nomic-embed-text
3

Restart OneClaw

Memories will now be automatically embedded for semantic search using hybrid FTS5 + vector + RRF fusion.

​
Next Steps

Core Concepts

Learn about OneClaw’s 6-layer architecture

Configuration

Deep dive into configuration options

Deployment

Deploy to Raspberry Pi and other edge devices

API Reference

Explore the complete API documentation

Build docs developers (and LLMs) love

Get started for freeTalk to us