OpenClaw CLI: Essential Commands and Options

Q: Output Formats

Many commands support --json for machine-readable output. This is useful for scripting:

Clawpedia · For Humans

A comprehensive reference for the OpenClaw command-line interface, including all flags and configuration options.

Overview

The OpenClaw CLI (Command-Line Interface) is the primary tool for managing every aspect of your agent. This comprehensive reference covers all available commands, their options, flags, environment variables, and output formats — everything a power user needs for full control.

Global Flags

These flags can be applied to any OpenClaw command:

Flag	Short	Description

`--config <path>`	`-c`	Use a custom config file instead of the default

`--verbose`	`-v`	Enable verbose output

`--quiet`	`-q`	Suppress non-essential output

`--json`	`-j`	Output in JSON format (useful for scripting)

`--no-color`		Disable colored terminal output

`--version`	`-V`	Print the OpenClaw version and exit


# Example: get the version
openclaw --version

# Example: list skills in JSON format
openclaw skills list --json

Environment Variables

`--help`	`-h`	Show help for any command

OpenClaw respects several environment variables that override config file values:

Variable	Description	Default

`OPENCLAW_HOME`	Base directory for all OpenClaw data	`~/.openclaw`

`OPENCLAW_CONFIG`	Path to the config file	`$OPENCLAW_HOME/config.yaml`

`OPENCLAW_LOG_LEVEL`	Log verbosity: `debug`, `info`, `warn`, `error`	`info`

`OPENCLAW_MODEL`	Override the default model	Config value

`OPENCLAW_PROVIDER`	Override the default provider	Config value

`OPENCLAW_API_KEY`	API key for the active provider	Config value

`OPENCLAW_PORT`	API server port	`3120`


# Start with a custom home directory
OPENCLAW_HOME=/opt/openclaw openclaw start

# Override model for a single session
OPENCLAW_MODEL=anthropic/claude-3 openclaw chat

`OPENCLAW_NO_TELEMETRY`	Disable anonymous usage telemetry	`false`

Tip: Environment variables take precedence over config file values but are overridden by explicit CLI flags.

Command Reference

`openclaw start`

Starts the OpenClaw agent.


openclaw start [flags]

Flag	Description

`--daemon`	Run as background process

`--port <number>`	Override API port

`--no-skills`	Start without loading skills

`--no-memory`	Start without loading persistent memory

`openclaw stop`

`--profile <name>`	Use a named configuration profile

Stops a running agent.


openclaw stop [flags]

Flag	Description

`--force`	Kill immediately without graceful shutdown

`openclaw restart`

`--timeout <seconds>`	Max wait time for graceful shutdown (default: 30)

Stops and restarts the agent.


openclaw restart [flags]

Accepts all flags from both start and stop.

`openclaw status`

Shows the current state of the agent.


openclaw status [flags]

Flag	Description

`--detail`	Show extended information (skills, memory, platforms)

`openclaw config`

`--json`	Output in JSON format

Manage configuration values.


openclaw config <subcommand> [args]

Subcommand	Description

`list`	Show all configuration values

`get <key>`	Get a specific value

`set <key> <value>`	Set a value

`reset <key>`	Reset a key to its default

`validate`	Check the config file for syntax errors

`edit`	Open the config file in your default editor


# Validate your config after manual edits
openclaw config validate

# Open in editor
openclaw config edit

`openclaw skills`

`export`	Export the full config to stdout

Manage agent skills.


openclaw skills <subcommand> [args]

Subcommand	Description

`list`	List all installed skills

`install <name>`	Install a skill from ClawHub

`remove <name>`	Uninstall a skill

`enable <name>`	Enable a disabled skill

`disable <name>`	Disable a skill without removing it

`update [name]`	Update a specific skill or all skills

`info <name>`	Show detailed information about a skill


# Install and immediately enable
openclaw skills install calendar-sync

# Check what a skill does before installing
openclaw skills info calendar-sync

`openclaw memory`

`create <name>`	Scaffold a new custom skill project

Manage the agent's persistent memory.


openclaw memory <subcommand> [args]

Subcommand	Description

`stats`	Show memory usage statistics

`search <query>`	Search memory entries

`export`	Export memory to JSON

`import <file>`	Import memory from a JSON file

`clear`	Delete all memory (requires `--confirm`)

`openclaw logs`

`prune`	Remove old or low-relevance entries

View agent logs.


openclaw logs [flags]

Flag	Description

`--follow` / `-f`	Stream logs in real time

`--last <n>`	Show the last n log entries

`--level <level>`	Filter by log level

`--since <duration>`	Show logs from the last duration (e.g., `1h`, `30m`)


# Stream errors only
openclaw logs --follow --level error

# Last hour of debug logs
openclaw logs --since 1h --level debug

`openclaw ask`

`--grep <pattern>`	Filter log lines matching a regex pattern

Send a one-shot prompt.


openclaw ask "<prompt>" [flags]

Flag	Description

`--model <name>`	Override the model for this request

`--no-memory`	Don't use or store memory for this query

`--system <prompt>`	Set a custom system prompt

`openclaw chat`

`--file <path>`	Attach a file for context

Start an interactive chat session.


openclaw chat [flags]

Flag	Description

`--model <name>`	Override the model

`--system <prompt>`	Set a system prompt

`openclaw doctor`

`--no-memory`	Disable memory for this session

Run diagnostic checks.


openclaw doctor [flags]

Flag	Description

`--platforms`	Test platform connections

`--keys`	Validate API keys

`--model`	Test model connectivity

`openclaw update`

`--all`	Run all checks

Update OpenClaw itself.


openclaw update [flags]

Flag	Description

`--check`	Check for updates without installing

Output Formats

`--pre-release`	Include pre-release versions

Many commands support --json for machine-readable output. This is useful for scripting:


# Parse status in a script
STATUS=$(openclaw status --json | jq -r '.state')
if [ "$STATUS" != "RUNNING" ]; then
  openclaw start --daemon
fi

Exit Codes

Code	Meaning

`0`	Success

`1`	General error

`2`	Configuration error

`3`	Connection error (platform or model)

`4`	Authentication error (invalid API key)

Shell Completion

`5`	Agent not running (when required)

OpenClaw supports autocompletion for Bash, Zsh, and Fish:


# Bash
openclaw completion bash >> ~/.bashrc

# Zsh
openclaw completion zsh >> ~/.zshrc

# Fish
openclaw completion fish > ~/.config/fish/completions/openclaw.fish

Troubleshooting

Command Not Found

Ensure OpenClaw is on your PATH. If installed via the install script:


export PATH="$HOME/.openclaw/bin:$PATH"

JSON Output Is Empty

Some commands only produce JSON output when the agent is running. Start the agent first.

Config Changes Not Taking Effect

Changes made with config set require a restart unless you pass --restart. Check with:


openclaw config get <key>  # Shows stored value
openclaw status --detail   # Shows active runtime value

Next Steps

Learn the most common commands in Basic Commands to Control Your OpenClaw Agent.
Configure your AI model with Setting Your OpenClaw AI Model and Provider.
Explore skill management in Introduction to OpenClaw Skills and Automation.

How to use the OpenClaw CLI (openclaw command) effectively? — Master the OpenClaw command-line interface with essential commands, flags, and productivity tips.
OpenClaw CLI Flags and Getting Command Help — Complete reference for all OpenClaw CLI flags, options, and how to access built-in command help.
Basic Commands to Control Your OpenClaw Agent — Master the essential commands to start, stop, configure, and interact with your OpenClaw agent.
Configuring OpenClaw for First Use — Essential configuration steps to get your OpenClaw agent running after installation, including API keys and preferences.
How to fix "openclaw command not recognized" in terminal? — Resolve the "command not recognized" error by checking your PATH, installation, and shell configuration.

OpenClaw CLI: Essential Commands and Options

Overview

Global Flags

Environment Variables

Command Reference

openclaw start

openclaw stop

openclaw restart

openclaw status

openclaw config

openclaw skills

openclaw memory

openclaw logs

openclaw ask

openclaw chat

openclaw doctor

openclaw update