Want to install and manage OpenClaw without repeatedly searching through documentation? Here’s a practical command reference covering installation, onboarding, configuration, models, channels, Gateway management and troubleshooting.

Commands may change between releases, so check your installed version with:

openclaw --version
openclaw --help
openclaw <command> --help

1. Install OpenClaw

macOS, Linux or WSL2:

curl -fsSL https://openclaw.ai/install.sh | bash

Windows PowerShell:

iwr -useb https://openclaw.ai/install.ps1 | iex

The installer detects the operating system, provisions Node.js when necessary, installs OpenClaw and launches onboarding.

To install without starting onboarding:

curl -fsSL https://openclaw.ai/install.sh | bash -s -- --no-onboard

Manual npm installation:

npm install -g openclaw@latest --allow-scripts=openclaw
openclaw onboard --install-daemon

OpenClaw currently requires Node.js 24.16+ or 26.1+, with Node 26 recommended. Full installation options are documented in the official installation guide.

2. First-time setup

Run the guided onboarding process:

openclaw onboard

Install OpenClaw’s managed Gateway service during onboarding:

openclaw onboard --install-daemon

Other useful setup options:

openclaw setup
openclaw setup --wizard
openclaw setup --classic
openclaw setup --baseline
openclaw setup --workspace ~/my-openclaw-workspace
openclaw setup --json

What they do:

• setup — chat-based setup assistant
• setup --wizard — complete interactive onboarding
• setup --classic — traditional multi-step wizard
• setup --baseline — create only the base configuration and workspace
• onboard — configure authentication, models, Gateway, workspace, channels and skills

Remote Gateway setup is also supported:

openclaw setup \
  --non-interactive \
  --accept-risk \
  --mode remote \
  --remote-url wss://gateway.example.com \
  --remote-token '<token>'

Never publish real API keys, passwords or Gateway tokens in screenshots, posts or shell history. The internet already contains enough avoidable excitement.

3. Reconfigure an existing installation

Launch the interactive configuration wizard:

openclaw configure

Configure only selected areas:

openclaw configure --section workspace
openclaw configure --section model
openclaw configure --section web
openclaw configure --section gateway
openclaw configure --section daemon
openclaw configure --section channels
openclaw configure --section plugins
openclaw configure --section skills
openclaw configure --section health

Multiple sections can be combined:

openclaw configure \
  --section model \
  --section channels \
  --section health

The available configuration sections and their behaviour are covered in the configure command reference.

4. Non-interactive configuration

Find the active configuration file:

openclaw config file

Read a setting:

openclaw config get gateway.port
openclaw config get agents.defaults.model --json

Change a setting:

openclaw config set gateway.port 18789
openclaw config set agents.defaults.heartbeat.every "2h"

Remove a setting:

openclaw config unset agents.defaults.fastModeDefault

Apply several changes from a JSON5 patch:

openclaw config patch \
  --file ./openclaw.patch.json5 \
  --dry-run

Validate before restarting anything:

openclaw config validate
openclaw config validate --json

Inspect the complete configuration schema:

openclaw config schema
openclaw config schema --json

Use --dry-run where supported before making important changes. Full examples are in the configuration CLI reference.

5. Configure models and providers

Inspect model status:

openclaw models status
openclaw models status --json

List available models:

openclaw models list
openclaw models list --provider ollama

Set the primary model:

openclaw models set openai/gpt-5.6-sol

Manage authentication:

openclaw models auth list
openclaw models auth login --provider openai

Manage fallbacks:

openclaw models fallbacks --help

Other model commands include:

openclaw models refresh
openclaw models aliases --help
openclaw models accounts --help
openclaw models set-image <provider/model>
openclaw models image-fallbacks --help

Use the models command reference for provider-specific authentication and model routing.

6. Optional local models with Ollama

Install Ollama, then download a model:

ollama pull gemma4

Run OpenClaw onboarding and select Ollama:

openclaw onboard

Verify model discovery:

openclaw models list --provider ollama

Make the local model your primary:

openclaw models set ollama/gemma4

OpenClaw’s onboarding supports local-only, cloud-only and combined Ollama configurations. See the official Ollama setup guide.

7. Configure messaging channels

List configured channels:

openclaw channels list

Show available and installable channels:

openclaw channels list --all

Add a channel using guided setup:

openclaw channels add

Examples:

openclaw channels login --channel whatsapp
openclaw channels add --channel telegram --token '***'

Inspect live channel health:

openclaw channels status
openclaw channels status --probe

Useful channel commands:

openclaw channels capabilities
openclaw channels logs --channel all
openclaw channels resolve --channel slack "#general"
openclaw channels dead-letters list

Channel configuration and troubleshooting are covered in the channels CLI guide.

8. Manage the Gateway

Check the service:

openclaw gateway status

Start, stop or restart it:

openclaw gateway start
openclaw gateway stop
openclaw gateway restart

Install or remove the managed service:

openclaw gateway install
openclaw gateway uninstall

Run it directly in the foreground:

openclaw gateway run

Replace an existing listener on the configured port:

openclaw gateway run --force

Test connectivity:

openclaw gateway health
openclaw gateway probe
openclaw gateway stability

Export support information:

openclaw gateway diagnostics

Discover Gateways on the network:

openclaw gateway discover

The Gateway is OpenClaw’s WebSocket service for channels, sessions, nodes and hooks. Its complete command set is documented in the Gateway CLI reference.

9. Health checks and diagnosis

Start with the quick checks:

openclaw health
openclaw status
openclaw gateway status

Run a detailed read-only diagnosis:

openclaw status --all
openclaw status --deep
openclaw status --usage
openclaw status --json

Check configuration and runtime health:

openclaw doctor
openclaw doctor --lint
openclaw doctor --lint --severity-min warning
openclaw doctor --deep
openclaw doctor --json

Apply recommended repairs only after reviewing the findings:

openclaw doctor --fix

Non-interactive repair:

openclaw doctor --fix --non-interactive

Additional diagnostic commands:

openclaw logs
openclaw channels logs --channel all
openclaw channels status --probe
openclaw models status
openclaw config validate
openclaw security --help
openclaw update --help

doctor can inspect configuration, model routing, plugins, skills, channels, Gateway services, local state and migrations. See the official Doctor guide.

10. The “something is broken” checklist

Run these in order:

openclaw config validate
openclaw health
openclaw gateway status
openclaw models status
openclaw channels status --probe
openclaw status --all
openclaw doctor --lint

If the findings recommend repairs:

openclaw doctor --fix
openclaw gateway restart
openclaw status --deep

That gives you a complete OpenClaw toolkit: installation, onboarding, targeted configuration, scripted configuration, local Ollama models, channel management, Gateway control and diagnostics.

Full CLI index:
https://docs.openclaw.ai/cli

#OpenClaw #Ollama #LocalAI #SelfHostedAI #OpenSourceAI #ArtificialIntelligence #Automation #Linux #DevOps #AIInfrastructure