Docs
** — Official documentation
Last verified:
What is Docs?
OpenClaw is a self-hosted gateway that connects messaging apps such as WhatsApp, Telegram, Discord, iMessage, Matrix, Signal, Slack, Microsoft Teams, Google Chat, and more to AI coding agents. It is designed to run as a single Gateway process on your own machine or server, acting as a bridge between your chat apps and an always-available AI assistant.
The product emphasizes broad platform support, with docs covering installation on macOS, Linux, Windows, Docker, Kubernetes, Nix, Raspberry Pi, cloud hosts, and more. It also supports browser control, remote access, Tailscale, and OpenAI-compatible HTTP endpoints, so it can be used both as a personal assistant and as an automation gateway.
OpenClaw includes a large CLI and configuration surface for onboarding, messaging, pairing, automation, cron jobs, background tasks, models, memory, sessions, sandboxing, secrets, and debugging. The docs show that it is built for power users who want to connect multiple channels, customize model providers, tune security and access controls, and automate workflows through hooks, plugins, and tools.
It is especially suited to people who want a self-hosted AI layer over their communication channels rather than a single chat app. Teams or solo operators who need control over authentication, routing, model selection, and extensibility would likely get the most value from it.
Docs pricing
Pricing model: Freemium
The docs do not present a public pricing page in the material reviewed, and they describe OpenClaw as a self-hosted product. Setup requires Node.js and an API key from a model provider such as Anthropic, OpenAI, or Google, and the docs mention configuration of model providers and pricing bootstrap, but no free tier, subscription tiers, or paid plan breakdown is stated on the website content reviewed.
Docs pros
- Self-hosted gateway architecture
- Supports many chat platforms
- Works across multiple operating systems
- Single Gateway process model
- Can bridge chats to AI agents
- Offers CLI onboarding and setup
- Includes browser control capabilities
- Supports remote access options
- Has Tailscale integration
- Provides OpenAI-compatible endpoints
- Highly configurable gateway auth
- Fine-grained channel access controls
- Supports plugins and extensions
- Supports MCP server definitions
- Supports custom model providers
- Includes memory and session management
- Has automation hooks and scheduled tasks
- Supports multiple gateways on one host
- Includes dashboard and Control UI
- Can customize the UI root and branding
Docs cons
- Requires self-hosting and setup
- Needs a model-provider API key
- Documentation is highly complex
- Large configuration surface
- Some features require restart
- Many capabilities depend on plugins
- Security options can be tricky
- Channel-specific setup can be time-consuming
- Browser and remote access settings can be delicate
- Not a plug-and-play consumer app
Frequently asked questions about Docs
What is OpenClaw?
OpenClaw is a self-hosted gateway for AI agents that connects chat apps like WhatsApp, Telegram, Discord, iMessage, Matrix, Slack, Signal, Microsoft Teams, Google Chat, and others to an AI assistant. You run one Gateway process on your machine or server, and it becomes the bridge between those channels and your AI workflows.
Which platforms does OpenClaw support?
The docs show support for a wide range of platforms, including macOS, Linux, Windows, Docker, Kubernetes, Nix, Raspberry Pi, and multiple cloud and VPS deployment options. The product also has dedicated docs for mobile and web surfaces such as Android, iOS, Linux app, macOS app, WebChat, and dashboard experiences.
How do you install OpenClaw?
The getting-started docs show quick install scripts for macOS/Linux and Windows PowerShell, plus other install methods such as Docker, Nix, and npm. After installation, onboarding walks you through selecting a model provider, setting an API key, and configuring the Gateway.
What do I need before onboarding?
You need Node.js, with Node 24 recommended and Node 22.19+ also supported, plus an API key from a model provider such as Anthropic, OpenAI, or Google. The onboarding flow prompts you for provider selection and key setup.
What does the Gateway do?
The Gateway is the core OpenClaw service that connects channels, tools, models, sessions, memory, and automation features. It listens on a local port by default, exposes a Control UI, and can integrate with browser control, MCP servers, plugins, and messaging channels.
Can OpenClaw work with custom models and providers?
Yes. The docs include a large provider directory and configuration reference for model providers, custom providers, allowlists, and local model services. OpenClaw also supports pricing bootstrap and provider-based model configuration.
Does OpenClaw support automation?
Yes. The documentation includes automation features such as hooks, scheduled tasks, standing orders, background tasks, task flow, and webhooks. These are intended for triggering agent runs and routing external events into the Gateway.
Can I extend OpenClaw with plugins or skills?
Yes. OpenClaw supports plugins, bundled and custom skills, plugin hooks, tool plugins, and plugin-specific configuration. The docs also cover plugin discovery, loading paths, manifests, and compatibility with other bundle formats.
Does OpenClaw have security controls?
Yes. The docs cover gateway authentication, trusted proxy auth, rate limiting, tool allow/deny controls, SSRF protections for browser usage, secrets management, and channel-level access controls. Many features are designed to be safe by default but configurable for trusted environments.
Is there a dashboard or UI?
Yes. The docs mention a dashboard and a browser-based Control UI that opens after the Gateway starts. There are also configuration options for customizing the UI root, assistant identity, and related browser and gateway behavior.