### OpenClaw Development Setup

Source: https://clawdocs.org/contributing/openclaw

Steps to clone the repository, install dependencies, build the project, run tests, and start the development server.

```bash
git clone https://github.com/openclaw/openclaw.git
cd openclaw

pnpm install

pnpm build

pnpm test

pnpm dev
```

--------------------------------

### Quick Start WebClaw Installation

Source: https://clawdocs.org/guides/webclaw

Use this command to quickly download and start WebClaw locally. For more advanced configuration, refer to the installation guide.

```bash
npx webclaw  

```

--------------------------------

### Install and Serve with vLLM (Single GPU)

Source: https://clawdocs.org/guides/local-models

Install vLLM and start a single-GPU server for a specified model. Ensure you have GPU hardware available.

```bash
pip install vllm  

# Single GPU  
vllm serve meta-llama/Llama-3.1-70B-Instruct \  
  --host 0.0.0.0 \  
  --port 8000
```

--------------------------------

### Clone and Start Local Development Server

Source: https://clawdocs.org/contributing/docs

Clone the repository, navigate to the directory, install dependencies, and start the local development server for hot-reloading.

```bash
git clone https://github.com/clawdocs/clawdocs.github.io.git
cd clawdocs
npm install
npm start
```

--------------------------------

### Install and Serve with SGLang

Source: https://clawdocs.org/guides/local-models

Install SGLang and launch a server for a specified model. This setup is optimized for high-throughput serving.

```bash
pip install sglang  

python -m sglang.launch_server \  
  --model meta-llama/Llama-3.1-70B-Instruct \  
  --port 30000
```

--------------------------------

### Install GitHub Copilot Plugin

Source: https://clawdocs.org/guides/plugin-system

Example of how to install the GitHub Copilot plugin using the OpenClaw CLI.

```bash
# Install the GitHub Copilot plugin  
openclaw plugins install @openclaw/copilot  
```

--------------------------------

### Install a Channel Plugin

Source: https://clawdocs.org/guides/channels

Install a new channel by adding its corresponding plugin, for example, the Matrix channel plugin.

```bash
# Install a plugin channel  
openclaw plugins install @openclaw/channel-matrix  
```

--------------------------------

### Install and Configure GitHub MCP Server

Source: https://clawdocs.org/guides/mcp-servers

Example of adding the GitHub MCP server, with an alternative manual configuration shown in comments.

```bash
# Install and configure the GitHub server
openclaw mcp add github
# Or manually:
# "github": { "command": "npx", "args": ["-y", "@modelcontextprotocol/server-github"], "env": {"GITHUB_PERSONAL_ACCESS_TOKEN": "${GITHUB_TOKEN}"}}

```

--------------------------------

### Run OpenClaw Onboarding Wizard

Source: https://clawdocs.org/getting-started/installation

Execute the onboarding wizard after installation to connect your LLM, install the gateway daemon, and configure messaging channels. The `--install-daemon` flag ensures the gateway starts on boot.

```bash
openclaw onboard --install-daemon  
```

--------------------------------

### Run Onboard Setup Wizard

Source: https://clawdocs.org/reference/cli

Initiates the interactive setup wizard for system configuration. Use flags to customize the setup process.

```bash
openclaw onboard [--install-daemon] [--skip-channels]  

```

--------------------------------

### Install and Serve with vLLM (Multi-GPU)

Source: https://clawdocs.org/guides/local-models

Install vLLM and start a multi-GPU server using tensor parallelism for a specified model. This requires multiple GPUs.

```bash
pip install vllm  

# Multi-GPU (tensor parallelism)  
vllm serve meta-llama/Llama-3.1-70B-Instruct \  
  --host 0.0.0.0 \  
  --port 8000 \  
  --tensor-parallel-size 2
```

--------------------------------

### List Installed Plugins

Source: https://clawdocs.org/guides/plugin-system

Example of how to check which plugins are currently installed in the OpenClaw system.

```bash
# Check what's installed  
openclaw plugins list  
```

--------------------------------

### Start OpenClaw Terminal UI

Source: https://clawdocs.org/guides/first-7-days/day-1-make-it-yours

Use the terminal UI to interact with your bot for setup and configuration. This is the recommended method for initial setup.

```bash
openclaw tui
```

--------------------------------

### Agent Setup and Gateway Start

Source: https://clawdocs.org/guides/multi-agent

Create agent workspaces, assign personalities using SOUL.md files, and start the OpenCLAW gateway to run all configured agents.

```bash
# Create workspaces
mkdir -p ~/.openclaw/workspace-assistant ~/.openclaw/workspace-coder

# Give each agent a personality
echo "You are a friendly personal assistant." > ~/.openclaw/workspace-assistant/SOUL.md
echo "You are a senior software engineer." > ~/.openclaw/workspace-coder/SOUL.md

# Start the gateway — both agents run on one process
openclaw start

```

--------------------------------

### Onboard OpenClaw Agent

Source: https://clawdocs.org/getting-started/quick-start

Run the onboard command to start the interactive setup wizard for your OpenClaw agent. This wizard guides you through selecting an LLM provider, entering API keys, and optionally connecting messaging channels.

```bash
openclaw onboard  

```

```text
🦞 Welcome to OpenClaw!  
  
? Choose your LLM provider:  
  ❯ Anthropic (Claude) — recommended  
    OpenAI (GPT)  
    Google (Gemini)  
    DeepSeek  
    OpenRouter (200+ models)  
    Local model (Ollama/LM Studio)  
  
? Enter your Anthropic API key: sk-ant-api03-****  
  
? Connect a messaging channel now?  
  ❯ Skip for now  
    WhatsApp  
    Telegram  
    Discord  
    Slack  
  
✓ Configuration saved to ~/.openclaw/openclaw.json  
✓ Run 'openclaw gateway' to start your agent  

```

--------------------------------

### Install Mission Control

Source: https://clawdocs.org/guides/multi-agent

Clone the Mission Control repository and start the Docker Compose services to set up the platform.

```bash
# Install  
git clone https://github.com/abhi1693/openclaw-mission-control  
cd openclaw-mission-control && docker compose up -d  

```

--------------------------------

### Start OpenClaw as an MCP Server

Source: https://clawdocs.org/guides/api-webhooks

Use this command to start OpenClaw and expose its installed skills as MCP tools over stdio or HTTP.

```bash
# Start OpenClaw as an MCP server  
openclaw mcp serve  
```

--------------------------------

### Install and Build WebClaw from Source

Source: https://clawdocs.org/guides/webclaw

Clone the repository, install dependencies, and build for development or production.

```bash
git clone https://github.com/ibelick/webclaw.git  
cd webclaw  

# Install dependencies  
pnpm install  

# Development mode (with hot reload)  
pnpm dev  

# Production build  
pnpm build  
pnpm start  

```

--------------------------------

### Add a New MCP Server Interactively

Source: https://clawdocs.org/guides/mcp-servers

Use this command to start an interactive setup process for a new MCP server.

```bash
# Add a new MCP server interactively
openclaw mcp add my-database
```

--------------------------------

### Install Skill from URL

Source: https://clawdocs.org/guides/clawhub

Install a skill directly from any Git repository by providing its URL.

```bash
# Install from any Git repository  
openclaw clawhub install https://github.com/user/skill-repo/skill.md  
```

--------------------------------

### Install PinchChat using Docker

Source: https://clawdocs.org/guides/webclaw

Clone the PinchChat repository and start the Docker containers to run the application.

```bash
git clone https://github.com/MarlBurroW/pinchchat.git
cd pinchchat
docker compose up -d
```

--------------------------------

### Install 1Panel

Source: https://clawdocs.org/guides/deployment-options

Execute this one-line command to install 1Panel on any Linux server. After installation, access the web panel via the provided URL and credentials.

```bash
bash -c "$(curl -sSL https://resource.fit2cloud.com/1panel/package/v2/quick_start.sh)"
```

--------------------------------

### Install and Configure OpenClaw

Source: https://clawdocs.org/

This snippet shows how to install OpenClaw using a curl script, configure an LLM provider and model, add a Telegram channel, and start the agent.

```bash
# Install OpenClaw
curl -fsSL https://openclaw.ai/install.sh | bash

# Configure your LLM provider
openclaw config set provider openrouter
openclaw config set model deepseek/deepseek-chat-v3-0324

# Connect Telegram
openclaw channel add telegram

# Launch
openclaw start
```

--------------------------------

### OpenClaw Setup Commands

Source: https://clawdocs.org/guides/basic-usage

Illustrates the sequence of commands for initial OpenClaw setup and configuration, including adding channels.

```bash
# Full setup from scratch  
openclaw setup  
openclaw channels add whatsapp  
openclaw onboard  

```

--------------------------------

### Install Software

Source: https://clawdocs.org/guides/first-7-days/day-2-daily-driver

Use this command to install a package (e.g., htop) if it is not already present on the system.

```shell
Install htop if it's not already installed  

```

--------------------------------

### Install BibiGPT Skill

Source: https://clawdocs.org/guides/video-generation

Install BibiGPT for AI-powered video summarization across various platforms, useful for research and content repurposing.

```bash
openclaw skills install bibigpt
```

--------------------------------

### Project Setup for Custom Webhook Channel

Source: https://clawdocs.org/guides/custom-channels

Initializes a new Node.js project, installs necessary dependencies including OpenClaw SDK and Express, and sets up TypeScript configuration.

```bash
mkdir openclaw-channel-webhook-custom
cd openclaw-channel-webhook-custom
npm init -y
npm install @openclaw/sdk express
npm install -D typescript @types/express @types/node
npx tsc --init

```

--------------------------------

### OpenClaw Gateway Output Example

Source: https://clawdocs.org/getting-started/quick-start

This is the expected output when the OpenClaw gateway starts successfully. It shows the version, control plane, and model details.

```text
🦞 OpenClaw Gateway v2026.6.1  
   WebSocket control plane: ws://localhost:18789  
   Heartbeat interval: 30m  
   Model: claude-sonnet-4-6  
   Status: RUNNING  


```

--------------------------------

### Mem0 Python SDK Example

Source: https://clawdocs.org/guides/memory-systems

Demonstrates basic usage of the Mem0 Python SDK for adding and searching memories. Requires the mem0 library to be installed.

```python
from mem0 import Memory
m = Memory()
m.add("User prefers TypeScript over JavaScript", user_id="alex")
results = m.search("What languages does the user prefer?", user_id="alex")

```

--------------------------------

### Install a Skill

Source: https://clawdocs.org/guides/first-7-days/day-4-skills

Use this command to install a skill from the clawhub. The output shows the download, verification, and installation process.

```bash
openclaw clawhub install <skill-name>
```

```text
Downloading skill: <skill-name>@1.2.0
Verifying VirusTotal scan... passed
Installing to ~/.openclaw/skills/<skill-name>.md
Done. Skill is active immediately — no restart needed.
```

--------------------------------

### Handle Gateway Startup Failures After Plugin Install

Source: https://clawdocs.org/guides/plugin-system

Disable a problematic plugin to allow the gateway to start, then use the doctor command to diagnose the issue.

```bash
openclaw plugins disable <name>   # Disable the problem plugin
openclaw gateway restart          # Start without it
openclaw plugins doctor           # Diagnose
```

--------------------------------

### Install ElevenLabs Package

Source: https://clawdocs.org/guides/video-generation

Install the Remotion ElevenLabs package for high-quality voiceover generation.

```bash
npm install @remotion/elevenlabs  

```

--------------------------------

### Install from npm

Source: https://clawdocs.org/guides/custom-channels

Commands to install and restart OpenCLAW after publishing to npm.

```bash
npm install @yourscope/openclaw-channel-webhook-custom
openclaw gateway restart
```

--------------------------------

### Install OpenClaw with Bun

Source: https://clawdocs.org/reference/resources

Install the OpenClaw package globally using Bun.

```bash
bun add -g openclaw
```

--------------------------------

### Install Skill from ClawHub

Source: https://clawdocs.org/guides/clawhub

Install the latest version of a skill or a specific version. You can also install with sandbox restrictions.

```bash
# Install latest version  
openclaw clawhub install email-triage  
  
# Install a specific version  
openclaw clawhub install email-triage@2.1.0  
  
# Install with sandbox restrictions  
openclaw clawhub install email-triage --sandbox --no-shell  
```

--------------------------------

### Build and Install Plugin Commands

Source: https://clawdocs.org/guides/plugin-system

Commands to build the plugin locally, install it for testing, or publish it to npm for wider distribution.

```bash
# Build
npm run build

# Install locally for testing
openclaw plugins install ./my-plugin

# Or publish and install from npm
npm publish --access public
openclaw plugins install @yourscope/my-plugin

```

--------------------------------

### Skill Installation Confirmation

Source: https://clawdocs.org/guides/first-7-days/day-6-power-moves

Confirms successful installation of the 'daily-standup' skill and lists its triggers.

```bash
✓ Installed "daily-standup" (v1.0.0)
  Triggers: standup, stand-up, daily standup, what did i do
```

--------------------------------

### Install OpenClaw with pnpm

Source: https://clawdocs.org/reference/resources

Install the OpenClaw package globally using pnpm.

```bash
pnpm add -g openclaw
```

--------------------------------

### Install Whisper Package

Source: https://clawdocs.org/guides/video-generation

Install the Remotion OpenAI Whisper package to enable caption generation from audio.

```bash
npm install @remotion/openai-whisper  

```

--------------------------------

### Install Email Assistant Skill

Source: https://clawdocs.org/guides/recipes/email-assistant

Installs the email assistant skill locally using its markdown file.

```bash
openclaw skill install ./email-assistant.md
```

--------------------------------

### Install OpenClaw via curl

Source: https://clawdocs.org/getting-started/quick-start

Use this command to download and run the OpenClaw installation script.

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

```

--------------------------------

### Create Prompt-to-Video Project from Template

Source: https://clawdocs.org/guides/video-generation

Use the official 'prompt-to-video' template to quickly set up a project that handles the full pipeline from text prompt to video generation.

```bash
# Create project from template  
npx create-video@latest --template prompt-to-video  
cd my-prompt-video  
  
# Set API keys  
export OPENAI_API_KEY=sk-...
export ELEVENLABS_API_KEY=...
  
# Generate a video from a prompt  
npm run gen -- "Explain how OpenClaw's heartbeat system works in 60 seconds"  

```

--------------------------------

### Install OpenClaw Observability Plugin

Source: https://clawdocs.org/guides/monitoring

Install the OpenClaw Observability Plugin to enable tracing and metrics export to OpenTelemetry backends.

```bash
# Install the plugin
openclaw plugins install openclaw-observability

```

--------------------------------

### Install Skills Safely

Source: https://clawdocs.org/security/skill-verification

Use these commands to install skills with enhanced security. The `--sandbox` flag prevents skills from escaping their environment, `--no-shell` restricts shell access, and `--to` allows manual review before installation.

```bash
# Install with sandbox (skills can't escape the sandbox)  
openclaw clawhub install <skill-name> --sandbox  

# Install but don't grant shell access  
openclaw clawhub install <skill-name> --no-shell  

# Install to a review directory first  
openclaw clawhub download <skill-name> --to ~/review/  
# ... review the files manually ...  
openclaw skill install ~/review/<skill-name>.md  

```

--------------------------------

### Search and Install ClawHub Skills

Source: https://clawdocs.org/guides/voice-multimodal

Use ClawHub to search for image generation skills and install specific ones like DALL-E.

```bash
# Browse image generation skills  
openclaw clawhub search "image generation"  
openclaw clawhub search "DALL-E"  
openclaw clawhub search "stable diffusion"  
  
# Install one  
openclaw clawhub install dalle-generator  
```

--------------------------------

### Install and Serve Model with vLLM

Source: https://clawdocs.org/guides/cloud-gpu-models

Install vLLM and serve a model using the command line. Ensure the model name and port are correctly specified.

```bash
pip install vllm

vllm serve Qwen/Qwen3-32B-Instruct \
  --host 0.0.0.0 \
  --port 8090 \
  --max-model-len 8192
```

--------------------------------

### Plugin Startup Sequence

Source: https://clawdocs.org/guides/plugin-system

Outlines the steps involved when the gateway starts and loads plugins.

```text
Gateway starts  
  → Load plugin manifests from install index (SQLite)  
  → Validate SecretRef contracts (required secrets present?)  
  → Instantiate plugins with config  
  → Execute startup hooks  
  → Register channels, providers, tools  
  → Ready to accept requests  

```

--------------------------------

### List Installed Plugins in JSON Format

Source: https://clawdocs.org/guides/plugin-system

Use the 'openclaw plugins list --json' command to get a full JSON dump of all installed plugins. This is useful for auditing and debugging plugin installations.

```bash
# Full JSON dump of installed plugins
openclaw plugins list --json
```

--------------------------------

### Diagnose Plugin Issues

Source: https://clawdocs.org/guides/plugin-system

Example of how to run the doctor command to diagnose potential issues with installed plugins.

```bash
# Something broken? Run doctor  
openclaw plugins doctor  
```

--------------------------------

### Smart Home Skill Usage Examples

Source: https://clawdocs.org/guides/recipes/smart-home

Illustrates how users can interact with the smart home skill using natural language commands for controlling lights and scenes.

```text
You: "Turn off the bedroom lights"
OpenClaw: Done. Bedroom lights are off.

You: "Set the office to bright cool white"
OpenClaw: Office lights set to 100% brightness, 6500K color temperature.
```

--------------------------------

### Create a New Remotion Project

Source: https://clawdocs.org/guides/video-generation

Use this command to initialize a new Remotion project. You can also use the '--template prompt-to-video' flag for a project pre-configured for prompt-based video generation.

```bash
npx create-video@latest my-video
```

```bash
npx create-video@latest --template prompt-to-video
```

--------------------------------

### Add Discord Channel to OpenClaw

Source: https://clawdocs.org/guides/first-7-days/day-5-multi-channel

Use this command to initiate the connection process for a new Discord channel. OpenClaw will guide you through credential setup.

```bash
openclaw channel add discord  
```

--------------------------------

### Set up Local Stable Diffusion with ComfyUI

Source: https://clawdocs.org/guides/voice-multimodal

Instructions for cloning the ComfyUI repository, installing dependencies, and running the local Stable Diffusion server.

```bash
# Install via Ollama (if supported) or standalone  
# Using ComfyUI for local Stable Diffusion  
git clone https://github.com/comfyanonymous/ComfyUI  
cd ComfyUI && pip install -r requirements.txt  
python main.py --listen 0.0.0.0 --port 8188  
```

--------------------------------

### Build a Custom Frontend with WebClaw

Source: https://clawdocs.org/guides/webclaw

Fork and clone the WebClaw repository, then navigate to the web client directory and start the development server.

```bash
# Fork and clone
git clone https://github.com/YOUR-USERNAME/webclaw.git
cd webclaw

# Work only on the web client
cd apps/webclaw
pnpm dev
```

--------------------------------

### Example Cron Jobs for Various Tasks

Source: https://clawdocs.org/guides/automation

Illustrates how to set up different cron jobs for daily standups, weekly reports, hourly news checks, and nightly backup verifications.

```bash
# Daily standup summary at 9 AM
openclaw cron add --name "Standup" --cron "0 9 * * 1-5" \
  --message "Summarize what I worked on yesterday from git logs and calendar."
  
# Weekly report every Friday at 5 PM
openclaw cron add --name "Weekly report" --cron "0 17 * * 5" \
  --message "Generate my weekly status report from this week's git activity."
  
# Hourly news check during work hours
openclaw cron add --name "News check" --cron "0 9-17 * * 1-5" \
  --message "Check for breaking news in AI and tech. Only alert if significant."
  
# Nightly backup verification
openclaw cron add --name "Backup check" --cron "0 2 * * *" \
  --message "Verify that nightly backups completed successfully."

```

--------------------------------

### Verify OpenClaw Setup Checklist

Source: https://clawdocs.org/getting-started/quick-start

Run these commands sequentially to check different aspects of your OpenClaw installation, from version to specific functionalities like chat and file access.

```bash
# 1. Check OpenClaw version  
openclaw --version  
```

```bash
# 2. Check gateway status  
openclaw status  
```

```bash
# 3. Send a test message  
openclaw chat "Hello! Tell me what you can do."  
```

```bash
# 4. Test file access  
openclaw chat "List the files in my current directory"  
```

```bash
# 5. Test shell execution  
openclaw chat "What's my current disk usage?"  
```

```bash
# 6. Check connected channels (if any)  
openclaw channels list  
```

```bash
# 7. Test heartbeat (if configured)  
openclaw heartbeat --dry-run  
```

--------------------------------

### View Skill Details on ClawHub

Source: https://clawdocs.org/guides/first-7-days/day-4-skills

Use the 'openclaw clawhub info' command followed by the skill name to get detailed information about a specific skill. This is essential before installing or using a skill.

```bash
openclaw clawhub info <skill-name>
```

--------------------------------

### Perform OpenCLAW Upgrade

Source: https://clawdocs.org/guides/cicd-testing

Outlines the steps for upgrading OpenCLAW, including checking release notes, updating the package, validating the installation, and restarting the gateway. Ensure a backup is performed before starting the upgrade.

```bash
# 1. Backup (see above)  
  
# 2. Check release notes  
gh release view --repo openclaw/openclaw  
  
# 3. Upgrade  
npm update -g openclaw  
  
# 4. Validate  
openclaw doctor  
  
# 5. Restart gateway  
openclaw gateway restart  
  
# 6. Verify channels reconnected  
openclaw channel list  

```

--------------------------------

### Create a Proposal with Support Files

Source: https://clawdocs.org/guides/skill-workshop

This command creates a proposal and includes additional support files like templates, scripts, and references from a specified directory. Constraints apply to the number and size of files.

```bash
openclaw skills workshop propose-create \
  --name "config-validator" \
  --proposal ./my-proposal/PROPOSAL.md \
  --proposal-dir ./my-proposal  

```

--------------------------------

### Heartbeat Setup Commands

Source: https://clawdocs.org/guides/heartbeat

Commands to create the heartbeat file, restart the gateway, and test the heartbeat functionality immediately or in dry-run mode.

```bash
# Create your heartbeat file  
vim ~/.openclaw/HEARTBEAT.md  
  
# Restart the gateway to pick up changes  
openclaw gateway restart  
  
# Test immediately (doesn't wait for timer)  
openclaw heartbeat --now  
  
# Dry run (shows what would happen)  
openclaw heartbeat --dry-run  
```

--------------------------------

### Install a Skill

Source: https://clawdocs.org/guides/skill-development

Shows the command to install a skill locally using `openclaw skill install`. Once installed, the skill is immediately available for use.

```bash
openclaw skill install ~/.openclaw/skills/git-summary.md

```

--------------------------------

### Create Code Hike Template

Source: https://clawdocs.org/guides/video-generation

Use the create-video CLI to initialize a project with the code-hike template for animated code explanations.

```bash
npx create-video@latest --template code-hike  

```

--------------------------------

### Manage Installed Skills

Source: https://clawdocs.org/reference/cli

Manages installed skills, supporting installation from a path, removal, listing, testing without installation, and configuration. Specific skill names or paths are often required.

```bash
openclaw skill <install|remove|list|test|config> [skill-name|path]  

```

--------------------------------

### Create Heartbeat Directory and File

Source: https://clawdocs.org/guides/first-7-days/day-3-automation

Creates the necessary directory and opens the heartbeat configuration file in nano. This is the first step to setting up the autonomous loop.

```bash
mkdir -p ~/.openclaw/workspace  
openclaw gateway restart  
```

--------------------------------

### Install OpenClaw via npm

Source: https://clawdocs.org/getting-started/installation

Install OpenClaw globally using npm. This method requires Node.js and npm to be installed.

```bash
npm i -g openclaw  
openclaw onboard  
```

--------------------------------

### Install MCP Knowledge Base Server (Supermemory)

Source: https://clawdocs.org/guides/recipes/advanced-recipes

Install the Supermemory MCP server using npx for managing larger knowledge bases.

```bash
# Or use Supermemory's MCP server
npx @supermemory/mcp-server
```

--------------------------------

### Multi-File Skill Structure Example

Source: https://clawdocs.org/guides/clawhub

Illustrates the directory structure for a multi-file skill, including the main SKILL.md, template files, helper scripts, and usage examples.

```text
skills/code-reviewer/  
├── SKILL.md           # Main skill definition  
├── templates/  
│   └── review.md      # Review template  
├── scripts/  
│   └── lint.sh        # Helper script  
└── examples/  
    └── sample.md      # Usage example  

```

--------------------------------

### List Installed Claw Skills

Source: https://clawdocs.org/guides/first-7-days/day-4-skills

View a list of all skills currently installed on your system. This is useful for keeping track of your installed skill set.

```bash
# List all installed skills
openclaw skill list
```

--------------------------------

### File-Based State Management Example

Source: https://clawdocs.org/guides/skill-development

This example outlines storing and updating skill state using Markdown files within the agent's memory directory. It details reading, updating, and writing state after each invocation.

```markdown
## State Management  
  
After each invocation:  
1. Read current state from `~/.openclaw/memory/skills/expense-tracker-state.md`  
2. Update with new data  
3. Write the updated state back  
  
State file format:  
- Running total at the top  
- Individual entries as a Markdown list with dates  
- Monthly subtotals  

```

--------------------------------

### Update OpenClaw Installation

Source: https://clawdocs.org/getting-started/installation

Update OpenClaw if it was installed via npm or the one-liner script. For Git installations, pull the latest changes and rebuild.

```bash
# If installed via npm/one-liner  
npm update -g openclaw  
  
# If installed via git  
cd openclaw  
git pull  
pnpm install  
pnpm build  
```

--------------------------------

### Install OpenClaw via One-Liner

Source: https://clawdocs.org/getting-started/installation

Use this command to quickly download and install the latest release of OpenClaw globally. It automates the installation process and prompts for onboarding.

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

--------------------------------

### Troubleshoot Whisper Installation and Usage

Source: https://clawdocs.org/guides/voice-multimodal

Verify Whisper is installed and test its basic functionality with a sample audio file. Also, check if ffmpeg, a required dependency, is installed.

```bash
# Check Whisper is installed  
whisper --help  
```

```bash
# Test with a sample file  
whisper --model base test.ogg  
```

```bash
# Check ffmpeg (required by Whisper)  
ffmpeg -version  
```

--------------------------------

### Generate Video with `video_generate` Tool

Source: https://clawdocs.org/guides/video-generation

Use the `video_generate` tool for quick video clips from text prompts. No additional setup is required.

```bash
openclaw chat "Generate a 5-second video of a hello world animation"  

```

--------------------------------

### Fix npm Global Install Permissions

Source: https://clawdocs.org/getting-started/quick-start

If you receive an 'EACCES permission denied' error during npm global installations, use this command to install packages with sufficient permissions.

```bash
sudo npm i -g openclaw
```

--------------------------------

### Starting the OpenClaw Gateway

Source: https://clawdocs.org/architecture/gateway

Demonstrates various ways to start the OpenClaw Gateway process, including development, production (daemon mode), custom ports, and debug logging.

```bash
openclaw gateway

```

```bash
openclaw gateway --daemon

```

```bash
openclaw gateway --port 19000

```

```bash
openclaw gateway --log-level debug

```

--------------------------------

### Full OpenClaw Configuration Example

Source: https://clawdocs.org/reference/configuration

A comprehensive example of the OpenClaw configuration file, demonstrating settings for LLM, models, gateway, execution environment, sandbox, tools, agents, bindings, session, DM access control, heartbeat, memory, skills, permissions, MCP servers, channels, and logging.

```json
{
  // LLM Configuration
  "brain": {
    "provider": "anthropic",
    "model": "claude-opus-4-6",
    "api_key": "${ANTHROPIC_API_KEY}",
    "temperature": 0.7,
    "max_tokens": 4096,
    "fallback": {
      "provider": "openrouter",
      "model": "deepseek-v3.2"
    },
    "heartbeat_override": {
      "provider": "anthropic",
      "model": "claude-haiku-4-5-20251001"
    }
  },

  // Model Providers
  "models": {
    "mode": "merge",
    "providers": {
      "ollama": {
        "baseUrl": "http://localhost:11434",
        "models": {
          "qwen3:32b": {
            "contextWindow": 32768,
            "maxTokens": 4096
          }
        }
      }
    }
  },

  // Gateway
  "gateway": {
    "host": "127.0.0.1",
    "port": 18789,
    "max_connections": 10,
    "log_level": "info",
    "auth": {
      "mode": "token",
      "token": "${OPENCLAW_AUTH_TOKEN}"
    },
    "trustedProxies": ["127.0.0.1"]
  },

  // Execution Environment
  "hands": {
    "shell": {
      "enabled": true,
      "timeout": 30000,
      "blocked_commands": ["rm -rf /", "sudo rm"]
    },
    "browser": {
      "enabled": true,
      "headless": true,
      "timeout": 60000,
      "allowed_domains": []
    },
    "filesystem": {
      "writable_paths": [],
      "blocked_paths": ["~/.ssh", "~/.gnupg"]
    }
  },

  // Sandbox
  "sandbox": {
    "mode": "non-main",
    "backend": "docker",
    "scope": "agent"
  },

  // Tools
  "tools": {
    "profile": "coding",
    "alsoAllow": ["lobster"],
    "exec": {
      "security": "allowlist",
      "ask": "on-miss"
    }
  },

  // Agents
  "agents": {
    "defaults": {
      "model": "claude-sonnet-4-6",
      "sandbox": { "mode": "non-main" }
    },
    "list": [
      {
        "id": "researcher",
        "model": "claude-opus-4-8",
        "workspace": "~/research"
      },
      {
        "id": "monitor",
        "model": "ollama/qwen3:14b"
      }
    ]
  },

  // Bindings (message routing)
  "bindings": [
    {
      "match": { "channel": "telegram", "peer": "boss" },
      "agentId": "researcher"
    }
  ],

  // Session
  "session": {
    "dmScope": "per-channel-peer"
  },

  // DM Access Control
  "dm": {
    "accessPolicy": "pairing",
    "pairing": {
      "maxPending": 3,
      "expireMinutes": 60
    }
  },

  // Heartbeat
  "heartbeat": {
    "enabled": true,
    "interval": 1800,
    "model": "claude-haiku-4-5-20251001",
    "isolatedSession": true,
    "lightContext": false,
    "ackMaxChars": 300,
    "showOk": false,
    "showAlerts": true,
    "quiet_hours": {
      "start": "22:00",
      "end": "07:00",
      "timezone": "America/Los_Angeles"
    },
    "dreaming": {
      "enabled": false,
      "idle_threshold": 3
    }
  },

  // Memory
  "memory": {
    "enabled": true,
    "path": "~/.openclaw/memory",
    "max_context_tokens": 2000,
    "auto_save": true
  },

  // Skills
  "skills": {
    "path": "~/.openclaw/skills",
    "allow_install": true,
    "allow_clawhub": true,
    "auto_install_deps": false
  },

  // Permissions
  "permissions": {
    "mode": "ask",
    "auto_approve": ["git status", "git diff", "ls", "cat"],
    "always_deny": ["rm -rf /", "sudo rm"]
  },

  // MCP Servers
  "mcp": {
    "servers": {
      "filesystem": {
        "command": "npx @anthropic/mcp-filesystem",
        "args": ["/home/user"]
      },
      "github": {
        "command": "npx @anthropic/mcp-github",
        "env": { "GITHUB_TOKEN": "${GITHUB_TOKEN}" }
      }
    }
  },

  // Channels
  "channels": {
    "telegram": {
      "enabled": true,
      "allowedContacts": ["user123"]
    }
  },

  // Logging
  "logging": {
    "level": "info",
    "path": "~/.openclaw/logs",
    "max_size": "10m",
    "max_files": 5,
    "redactSensitive": "tools",
    "audit": { "enabled": false }
  }
}

```

--------------------------------

### Create Audiogram Template

Source: https://clawdocs.org/guides/video-generation

Use the create-video CLI to quickly set up an audiogram project for podcast clips.

```bash
# Use the audiogram template  
npx create-video@latest --template audiogram  

```

--------------------------------

### Example SOUL.md Content

Source: https://clawdocs.org/guides/first-7-days/day-1-make-it-yours

A starter template for your SOUL.md file. Define your bot's name, core truths, boundaries, vibe, and expertise.

```markdown
# Nova  
  
I am a sharp, curious assistant who values clarity over formality.  
  
## Core Truths  
- Honesty first — I say "I don't know" when I don't know  
- I explain my reasoning, not just my conclusions  
- I match the user's energy — brief when they're brief, deep when they're deep  
  
## Boundaries  
- I never fabricate sources or citations  
- I flag when I'm uncertain vs. confident  
- I won't help with anything that could harm people  
  
## Vibe  
Conversational but precise. I use plain language, avoid jargon unless the user speaks it,  
and I'd rather give one clear answer than five hedged ones.  
  
## Expertise  
Software development, system administration, research, writing.  
I learn the user's stack and preferences over time.  

```

--------------------------------

### OpenCLaw Skill Installation Commands

Source: https://clawdocs.org/guides/skill-development

Commands for installing skills and ClawHub components.

```bash
openclaw skill install <path>
openclaw clawhub install <name>[@version]
openclaw clawhub install <name> --sandbox --no-shell
```

--------------------------------

### Install OpenAI Whisper

Source: https://clawdocs.org/guides/voice-multimodal

Install the OpenAI Whisper library using pip or conda.

```shell
# Install via pip  
pip install openai-whisper  

# Or with conda  
conda install -c conda-forge openai-whisper  

```