178,000+ stars | 29,000+ forks | 500+ contributors | 8,817+ commits | MIT License
Fastest-growing GitHub repo ever — 9K to 178K stars in 60 days (18x faster than Kubernetes)
The most comprehensive, curated collection of OpenClaw resources, hosting guides, cost comparisons, security hardening, skills, tutorials, and community links.
OpenClaw (formerly Clawdbot → Moltbot) is a free, open-source autonomous AI agent created by Peter Steinberger (@steipete). It runs on your own hardware, connects to 10+ messaging platforms (WhatsApp, Telegram, Slack, Discord, Signal, iMessage, Google Chat, Microsoft Teams, Matrix, Zalo), and orchestrates AI agent workflows with persistent memory and 24/7 operation.
- What is OpenClaw?
- Name History
- System Requirements
- Quick Start (1 Minute)
- Setup Methods (1-10 Minutes)
- Hosting Providers Comparison
- Master Cost Comparison Table
- AI Model API Costs
- Total Real-World Cost Examples
- Cloudflare Workers (Moltworker) Deep Dive
- DigitalOcean 1-Click Deploy
- Hetzner Setup Guide
- Hostinger VPS Setup
- Oracle Cloud Free Tier Setup
- Raspberry Pi Setup
- Docker Deployment
- Security & Hardening
- Configuration
- Integrations & Features
- Performance Benchmarks
- OpenClaw vs Claude Code
- Alternatives & Competitors
- Moltbook (AI Social Network)
- Skills & Plugins
- Browser Automation
- Real-World Use Cases
- Tutorials & Guides
- Video Tutorials
- Community
- Media Coverage
- Common Issues & Troubleshooting
- Cost Calculator & Optimization
- Key Contributors
- Related Repositories
- Kubernetes & Helm Charts
- Courses & Learning Platforms
- Enterprise Considerations
- China & Global Adoption
- Latest Releases
OpenClaw is an agentic AI interface that:
- Runs locally on your own hardware (Mac Mini, VPS, Raspberry Pi, or serverless)
- Connects to 10+ messaging platforms simultaneously
- Has persistent memory across sessions (saves files, breadcrumbs, chat histories)
- Handles tasks spanning hours or days without losing context
- Operates 24/7 autonomously focusing on automation and integration
- Supports 5,700+ community-built skills via ClawHub
- Works with multiple AI providers (Anthropic, OpenAI, Google, OpenRouter, DeepSeek, local LLMs)
- Features voice assistant, browser automation, home automation, and cron scheduling
| Layer | Purpose |
|---|---|
| Gateway | Central control plane — message routing, sessions, plugins, tool execution policy |
| Channels | Adapters normalizing Telegram/WhatsApp/Discord/iMessage into standard message shapes |
| Routing + Sessions | Determines which agent handles specific conversations |
| Agent Runtime | Processes context, calls model providers, streams responses, requests tools |
| Tools | Capabilities — web fetch, browser control, command execution, device pairing |
| Surfaces | Interaction points — chat apps, web dashboard, macOS menu bar, Live Canvas |
| Date | Name | Reason |
|---|---|---|
| Nov 2025 | Clawdbot | Original name at launch |
| Jan 27, 2026 | Moltbot | Anthropic trademark complaint (similarity to "Claude") |
| Jan 30, 2026 | OpenClaw | Final rebrand, community vote |
When the team dropped the @clawdbot Twitter handle to transition to @moltbot, professional "handle snipers" immediately grabbed it. Scammers used the hijacked account to launch a fake CLAWD token on Solana.
| Component | Requirement |
|---|---|
| OS | macOS 11+ / Ubuntu 22.04+ / Windows (WSL2) |
| Runtime | Node.js 22+ |
| RAM | 2 GB (may crash during onboarding) |
| CPU | 2 cores |
| Storage | 10 GB SSD |
| Network | Stable internet connection |
| Component | Requirement |
|---|---|
| RAM | 4-8 GB (cloud models) / 16-24 GB (local LLMs) |
| CPU | 2-4 cores |
| Storage | 30-50 GB SSD |
| GPU | 12-24 GB VRAM (for local model inference) |
npm install -g openclaw@latest
openclaw onboard --install-daemon
openclaw gateway --port 18789 --verboseThat's it. The wizard walks you through API key setup, channel configuration, and security settings.
| Method | Time | Difficulty | Best For |
|---|---|---|---|
| Agent37 | 30 sec | Easy | Cheapest managed ($0.99/mo) |
| SimpleClaw | 1 min | Easy | Non-technical users |
| npm install | 1 min | Easy | Developers with Node.js |
| DigitalOcean 1-Click | 2 min | Easy | Quick cloud deploy |
| Railway Template | 2 min | Easy | PaaS users |
| Zeabur | 1 min | Easy | Docker auto-deploy |
| Docker | 5 min | Medium | Isolation & reproducibility |
| Cloudflare Workers | 5 min | Medium | Serverless enthusiasts |
| xCloud Managed | 5 min | None | Full managed hosting |
| Molty by Finna | 30 sec | None | Multiple Molties, Mission Control, GitHub sync |
| Manual VPS | 10 min | Medium | Full control |
| Raspberry Pi | 10 min | Medium | Low-power, always-on |
curl -fsSL https://openclaw.ai/install.sh | bash
openclaw onboard --install-daemonnpm install -g openclaw@latest
# or
pnpm add -g openclaw@latestgit clone https://github.com/openclaw/openclaw.git
cd openclaw
./docker-setup.sh| Platform | Link | Notes |
|---|---|---|
| DigitalOcean | 1-Click Deploy | Security-hardened, pre-configured |
| Railway | Template | Web-based /setup wizard |
| Render | render.yaml | Infrastructure as Code |
| SimpleClaw | simpleclaw.com | Deploy in under 1 minute |
| Zeabur | Template | One-click Docker deploy |
| Northflank | Stack | No server-side terminal needed |
| Lightning.AI | Environment | Browser-based, no local hardware |
| Coolify | Template | Self-hosted PaaS template |
| Elestio | Open Source | Fully managed in < 3 min |
| Provider | Free Resources | Limits | Permanent? | Setup Time | Best For |
|---|---|---|---|---|---|
| Oracle Cloud | 4 ARM CPUs, 24 GB RAM, 200 GB storage, 10 TB/mo | ARM architecture only | Yes (forever) | ~3 hours | Power users wanting $0 hosting |
| AMD Developer Cloud | MI300X GPU (192 GB memory), $100 credits | ~50 hours compute | No (credits) | 30 min | GPU-accelerated local LLMs |
| Google Cloud Run | 180K vCPU-sec, 360K GiB-sec, 2M requests/mo | Cold starts, stateless | Yes | 30 min | Serverless testing |
| Azure Container Apps | 180K vCPU-sec, 360K GiB-sec, 2M requests/mo | Cold starts | Yes | 30 min | Enterprise users |
| Render | Web service (spins down after 15 min) | Slow wake-up (~60s) | Yes | 5 min | Testing only |
| AWS Free Tier | t2.micro (1 vCPU, 1 GB RAM) | 12 months only | No (12 mo) | 15 min | AWS-familiar users |
| Railway | $5 one-time credit | 30 days trial | No | 2 min | Quick testing |
| Provider | Plan | Price/mo | vCPU | RAM | Storage | Bandwidth | Region |
|---|---|---|---|---|---|---|---|
| LumaDock | Basic | $1.99 | 1 | 1 GB | 20 GB SSD | 1 TB | US/EU |
| Vultr | Regular Cloud | $2.50 | 1 | 512 MB | 10 GB | 0.5 TB | 32 locations |
| Vultr | Standard | $3.50 | 1 | 1 GB | 25 GB NVMe | 1 TB | 32 locations |
| Hetzner | CX22 | €3.79 (~$4.15) | 2 | 4 GB | 40 GB SSD | 20 TB | EU (DE/FI) |
| Alibaba Cloud | Simple App Server | $4 | 1 | 1 GB | Varies | Varies | 19 regions |
| Contabo | Cloud VPS S | $4.95 | 4 | 8 GB | 50 GB SSD | Unlimited | US/EU/Asia |
| Hostinger | KVM 1 | $4.99 | 1 | 4 GB | 50 GB NVMe | 4 TB | Global |
| Linode | Shared 1GB | $5 | 1 | 1 GB | 25 GB | 1 TB | Global |
| Vultr | High-Performance | $6 | 1 | 1 GB | 25 GB NVMe | 2 TB | 32 locations |
| DigitalOcean | Basic Droplet | $6 | 1 | 1 GB | 25 GB | 1 TB | Global |
| Hetzner | CX32 | €6.80 (~$7.45) | 4 | 8 GB | 80 GB SSD | 20 TB | EU (DE/FI) |
| Contabo | Cloud VPS M | $7.95 | 4 | 8 GB | 200 GB SSD | Unlimited | US/EU/Asia |
| Kamatera | Custom | $4+ | Custom | Custom | Custom | Pay-as-you-go | Global |
| Zap-Hosting | VPS | Varies | Varies | Varies | Varies | Varies | EU |
| BoostedHost | OpenClaw VPS | Varies | Varies | Varies | Varies | Varies | Global |
| Provider | Plan | Price/mo | vCPU | RAM | Storage | Notes |
|---|---|---|---|---|---|---|
| DigitalOcean | Premium Droplet | $7 | 1 | 1 GB | 25 GB NVMe | 1-Click Deploy available |
| DigitalOcean | Standard 2GB | $12 | 1 | 2 GB | 50 GB | Recommended for OpenClaw |
| Hostinger | KVM 2 | $6.99 | 2 | 8 GB | 100 GB NVMe | Best Hostinger value |
| GCP | e2-medium | ~$12 | 2 | 4 GB | 30 GB SSD | $300 free credit (90 days) |
| Azure | B1ms VM | ~$15 | 1 | 2 GB | 32 GB | Enterprise compliance |
| Linode | Dedicated 4GB | $36 | 2 (dedicated) | 4 GB | 40 GB | Consistent performance |
| Platform | Starting Price | Free Tier? | Persistent Storage | Auto-Scale | Notes |
|---|---|---|---|---|---|
| Cloudflare Workers | $5/mo (paid plan) | 100K req/day (free) | R2 ($0.015/GB/mo) | Yes | Moltworker PoC |
| Railway | $5/mo (Hobby) | $5 credit trial | Yes (volumes) | Yes | Best PaaS UX |
| Render | $7/mo (web service) | Free w/ limits | Yes (disks) | Yes | YAML IaC |
| Fly.io | Pay-as-you-go | None | Yes (volumes) | Yes | Global edge |
| AWS Lightsail | $3.50/mo | None | Yes | No | Simple AWS VPS |
| Northflank | Varies | Limited | Yes | Yes | Stack templates |
| Zeabur | Varies | Limited | Yes | Yes | One-click Docker |
| Elestio | Varies | None | Yes | Yes | Fully managed open source |
| Coolify | Self-hosted | Free (self-host) | Yes | No | Open-source PaaS |
These providers handle ALL the setup for you — no Docker, no terminal, no DevOps required.
| Provider | Price/mo | Setup Time | Specs | What's Included | Promo |
|---|---|---|---|---|---|
| Agent37 | $0.99-3.99 | 30 sec | 2 vCPU, 4-6 GB RAM | Isolated instance, full UI, terminal, SSL | - |
| MyClaw.ai | $9 (was $29) | Instant | 2 vCPU, 4 GB RAM, 40 GB SSD | Auto-updates, backups, web terminal | 69-76% off early bird |
| get-open-claw.com | $9-49 | < 1 min | Varies by plan | OpenClaw Secure, daily backups, health monitoring | Pro includes $10 AI credits |
| EasyClaw | $10+ | 60 sec | Varies | Multi-model, 3+ platforms, zero maintenance | 29 min saved per deploy |
| ClawSimple | $8.25-29.08 | 2-3 min | Varies | Secure cloud, one-click updates (coming) | 20% off with LAUNCH20 |
| xCloud | $24 | 5 min | Managed | Telegram/WhatsApp pre-configured, encrypted backups | 7-day guarantee |
| ClawCloud | $29-129 | < 1 min | 1-2 vCPU, 1-4 GB RAM | All AI models (BYOK), auto-updates, daily backups | 70% off with EARLYBIRD70 |
| SimpleClaw | TBD | < 1 min | Managed | 1-click deploy, model selection | - |
| OpenClaw Cloud | Beta pricing | 5 min | Managed | All AI tokens included, 99.9% uptime, $0 hidden fees | Waitlist open |
| OpenClawd.ai | TBD | Minutes | Managed | Fully managed, security built-in | - |
| Kilo Claw | Pay-as-you-go | < 60 sec | Managed | 500+ models, 50+ platforms, SSO, audit logs | 7 days free compute |
| OpenClaw Hosting | $29+ | 5 min | Managed | Solo/Team/Business tiers, annual billing saves 20% | - |
| Myclawhost | $9-39 | Instant | Managed | Lite ($9), Pro ($19), Max ($39), full lifecycle | One-click deploy |
| Contabo OpenClaw | $4.95+ | Minutes | VPS | Unlimited workflows, full data ownership, predictable pricing | - |
| Molty by Finna | $19-99 | 30 sec | VM-isolated (Fly.io) | Multiple Molties per account, Mission Control (multi-agent coordination), GitHub backup sync, browser automation, auto-updates | 3-day free trial |
Hire someone to set it up for you.
| Provider | Price | Type | What's Included |
|---|---|---|---|
| GetClawHelp | $119-229 (one-time) | 1-on-1 video call | VPS setup, LLM config, 10-25 skills, 24/7 uptime |
| GetClawHelp Maintenance | $97/mo | Monthly | Updates, troubleshooting, new skills |
| Fiverr - almaasparvez | $90 | One-time | OpenClaw on AWS VPS, Telegram setup |
| Fiverr - marcos_mark683 | $40 | One-time | OpenClaw installation |
| Upwork - Custom Skills | Varies | One-time | Up to 23 custom skills, testing, step-by-step guide |
| Upwork - Secure Deploy | Varies | One-time | Mac Mini/VPS, sandboxing, human-in-the-loop approval gates |
| Freelancer.com | Varies | One-time | Linux setup, tuning, README |
| OpenClaw Money Playbook | $9.95 | E-book | Guide on monetizing OpenClaw setup services |
Business Model Insight: "The move right now is doing free OpenClaw installs, upselling security/skill packages/custom builds. We still make money on a 'free' install because we get an affiliate commission from the hosting company." — @GanimCorey on X
| Provider | Monthly Cost | Setup Time | Difficulty | 1-Click? | Best For |
|---|---|---|---|---|---|
| Oracle Cloud | $0 | 3 hours | Hard | No | Free-forever hosting |
| AMD Dev Cloud | $0 (credits) | 30 min | Medium | No | Free GPU inference |
| Agent37 | $0.99 | 30 sec | None | Yes | Cheapest managed |
| LumaDock | $1.99 | 15 min | Medium | No | Cheapest VPS |
| Vultr | $2.50 | 10 min | Medium | Yes | Global presence |
| Hetzner CX22 | $4.15 | 10 min | Medium | No | Best price/performance |
| Contabo VPS S | $4.95 | 15 min | Medium | No | Unlimited bandwidth |
| Hostinger KVM 1 | $4.99 | 10 min | Easy | Yes | Docker templates |
| Cloudflare Workers | $5 | 5 min | Medium | No | Serverless |
| Railway Hobby | $5 | 2 min | Easy | Yes | Best PaaS experience |
| Linode 1GB | $5 | 10 min | Medium | No | Consistent performance |
| DigitalOcean | $6 | 2 min | Easy | Yes | Best docs + 1-Click |
| Render | $7 | 5 min | Easy | Yes | YAML Infrastructure |
| MyClaw.ai Lite | $9 | Instant | None | Yes | Budget managed |
| DigitalOcean 2GB | $12 | 2 min | Easy | Yes | Recommended production |
| xCloud Managed | $24 | 5 min | None | Yes | Full managed hosting |
| ClawCloud Starter | $29 | < 1 min | None | Yes | Premium managed |
| Molty by Finna | $19 | 30 sec | None | Yes | Multi-Molty, Mission Control, GitHub sync |
| Raspberry Pi 5 | $0/mo | 30 min | Medium | No | Low-power, always-on |
| Mac Mini | $0/mo | 10 min | Easy | No | Privacy-first, local |
The real cost of running OpenClaw is the AI model API, not infrastructure.
| Provider | Model | Input Cost | Output Cost | Notes |
|---|---|---|---|---|
| Anthropic | Claude 3.5 Haiku | $0.80 | $4.00 | Best budget option |
| Anthropic | Claude 3.5 Sonnet | $3.00 | $15.00 | Best balance |
| Anthropic | Claude Opus 4.5 | $15.00 | $75.00 | Most capable, expensive |
| OpenAI | GPT-4o | $2.50 | $10.00 | Good alternative |
| OpenAI | GPT-4o-mini | $0.15 | $0.60 | Very cheap |
| Gemini 2.0 Flash | $0.10 | $0.40 | Cheapest hosted | |
| Gemini Flash-Lite | Free tier | Free tier | Zero cost option | |
| DeepSeek | DeepSeek V3 | $0.27 | $1.10 | Best value reasoning |
| Moonshot | Kimi K2.5 | $3.00 | $3.00 | Great agentic performance |
| Grok | Grok 4.1 mini | $0.20 | $0.50 | Budget alternative |
| OpenRouter | Various | Varies | Varies | Unified API for 200+ models |
| Ollama/Local | Any | $0 | $0 | Requires hardware (16 GB+ RAM) |
75x price difference between the most expensive (Claude Opus) and cheapest (Gemini Flash) options.
| Usage Level | Description | Monthly Cost |
|---|---|---|
| Minimal | Few messages/day, Gemini free tier | $0-5 |
| Light | ~50 messages/day, Claude Haiku | $5-15 |
| Moderate | ~200 messages/day, mixed models | $15-50 |
| Heavy | 500+ messages/day, Claude Sonnet | $50-150 |
| Power User | All-day usage, Claude Opus | $150-500+ |
- Use smart model routing — cheap models (Haiku/Flash) for simple tasks, expensive (Sonnet/Opus) for complex
- Enable prompt caching — Anthropic auto-applies 5-min cache, reducing repeat costs
- Use local LLMs via Ollama for zero API cost (requires 16 GB+ RAM)
- Set budget alerts in your AI provider dashboard
- Use OpenClaw Cost Calculator to estimate your spend
| Setup | Infrastructure | API | Total/month | Notes |
|---|---|---|---|---|
| Free Tier Max | Oracle Cloud ($0) | Gemini free ($0) | $0 | Limited but functional |
| Cheapest Managed | Agent37 ($1) | Haiku ($5) | $6 | 30-second setup |
| Ultra Budget | LumaDock ($2) | Haiku ($5) | $7 | Basic personal assistant |
| Budget Managed | MyClaw.ai Lite ($9) | BYOK ($10) | $19 | Zero setup, instant access |
| Sweet Spot | Hetzner CX22 ($4) | Mixed models ($15) | $19 | Best value-to-capability |
| Standard | DigitalOcean ($12) | Sonnet ($40) | $52 | Production-ready |
| Managed Easy | xCloud ($24) | Mixed ($30) | $54 | Zero technical setup |
| Cloudflare | Workers ($5+$30) | Mixed ($20) | $55 | Serverless architecture |
| Local LLM | Raspberry Pi 5 ($0/mo) | Ollama ($0) | $0 | After $80 hardware purchase |
| Power User | DigitalOcean ($24) | Opus ($200) | $224 | Heavy professional use |
| Extreme | Dedicated ($50) | All models ($573) | $623 | One developer's real report |
Moltworker is Cloudflare's official adaptation of OpenClaw for Workers.
- Container: Sandbox (standard-1: 1/2 vCPU, 4 GiB RAM, 8 GB disk)
- Storage: R2 for persistence (auto-sync every 5 minutes)
- Auth: Cloudflare Access (Zero Trust) + gateway token + device pairing
- Status: Proof-of-concept (not an official Cloudflare product)
git clone https://github.com/cloudflare/moltworker.git
cd moltworker && npm install
npx wrangler secret put ANTHROPIC_API_KEY
export MOLTBOT_GATEWAY_TOKEN=$(openssl rand -hex 32)
npx wrangler secret put MOLTBOT_GATEWAY_TOKEN
npm run deployAccess: https://your-worker.workers.dev/?token=YOUR_GATEWAY_TOKEN
| Component | Monthly Cost |
|---|---|
| Workers Paid Plan | $5.00 |
| Memory (~4 GiB) | ~$26.00 |
| CPU | ~$2.00 |
| Disk (8 GB) | ~$1.50 |
| Total | ~$34.50 |
Set
SANDBOX_SLEEP_AFTER=10mfor infrequent use to reduce costs significantly.
- Cannot support Telegram bots (requires persistent connection for long-polling)
- Containers hibernate when not receiving HTTP requests
- Use VPS deployment for Telegram integration
The fastest path to a production OpenClaw instance.
- Go to DigitalOcean Marketplace
- Click "Create OpenClaw Droplet"
- Choose $12/mo plan (2 GB RAM recommended)
- Select region, SSH key
- Deploy (ready in ~90 seconds)
Pre-configured: OpenClaw 2026.1.24-1, Docker isolation, unique gateway token, security-hardened, Caddy HTTPS.
- DigitalOcean Blog - Introducing OpenClaw
- Technical Deep Dive - Security Hardening
- Tutorial - How to Run OpenClaw
Best price-to-performance ratio for OpenClaw hosting.
| Plan | Price/mo | vCPU | RAM | Storage | Bandwidth |
|---|---|---|---|---|---|
| CX22 | €3.79 | 2 | 4 GB | 40 GB SSD | 20 TB |
| CX32 | €6.80 | 4 | 8 GB | 80 GB SSD | 20 TB |
| CAX11 (ARM) | €3.79 | 2 | 4 GB | 40 GB | 20 TB |
ssh root@your-server-ip
curl -fsSL https://deb.nodesource.com/setup_22.x | bash -
apt-get install -y nodejs
npm install -g openclaw@latest
openclaw onboard --install-daemon
ufw allow ssh && ufw allow 443/tcp && ufw enable| Plan | Price/mo | vCPU | RAM | Storage | Bandwidth |
|---|---|---|---|---|---|
| KVM 1 | $4.99 | 1 | 4 GB | 50 GB NVMe | 4 TB |
| KVM 2 | $6.99 | 2 | 8 GB | 100 GB NVMe | 8 TB |
| KVM 4 | $10.99 | 4 | 16 GB | 200 GB NVMe | 16 TB |
Renewal prices are ~2x. Lock in annual pricing.
Truly free forever — 4 ARM cores, 24 GB RAM, 200 GB storage, 10 TB/month.
curl -fsSL https://deb.nodesource.com/setup_22.x | bash -
apt-get install -y nodejs git
npm install -g openclaw@latest
openclaw onboard --install-daemonCaveats: ARM architecture, high demand (keep retrying), add credit card to prevent account termination.
| Device | Price | RAM | Performance |
|---|---|---|---|
| Pi 5 (8 GB) | $80 | 8 GB | Best choice |
| Pi 5 (4 GB) | $60 | 4 GB | Good budget option |
| Pi 4 (8 GB) | $55 | 8 GB | 2-2.5x slower than Pi 5 |
curl -fsSL https://deb.nodesource.com/setup_22.x | sudo bash -
sudo apt-get install -y nodejs
npm install -g openclaw@latest
openclaw onboard --install-daemongit clone https://github.com/openclaw/openclaw.git
cd openclaw && ./docker-setup.shdocker compose run --rm openclaw-cli pairing approve telegram <CODE>
docker compose run --rm openclaw-cli dashboard --no-open
docker compose run --rm openclaw-cli security audit --deep- Docker - OpenClaw Docs
- Running OpenClaw in Docker - Simon Willison
- Docker Guide for Beginners - Medium
| CVE | Severity | Description | Fixed In |
|---|---|---|---|
| CVE-2026-24763 | High | Docker PATH Command Injection | v2026.1.29 |
| GHSA-g8p2-7wf7-98mq | Critical (8.8) | gatewayUrl Token Exfiltration (1-click RCE) | v2026.1.29 |
| GHSA-q284-4pvr-m585 | High | sshNodeCommand Injection | v2026.1.29 |
| CVE-2026-25593 | Critical | Unauthenticated Local RCE via WebSocket | v2026.1.30 |
| CVE-2026-25475 | High | Local File Inclusion via MEDIA: Path | v2026.1.30 |
| CVE-2026-25253 | Critical (8.8) | 1-Click RCE via Cross-Site WebSocket Hijacking | v2026.1.29 |
Update immediately to v2026.2.1+ to patch all known vulnerabilities.
- Bind to
127.0.0.1only (never0.0.0.0) — over 30,000 exposed instances found online - Use SSH tunnels or Tailscale Serve for remote access
- Use auth profiles (system keychain):
openclaw auth set ANTHROPIC_API_KEY - Configure firewall:
ufw default deny incoming && ufw allow ssh && ufw allow 443/tcp && ufw enable - Use Docker for container isolation
- Run regular audits:
openclaw security audit --deepandopenclaw doctor - Use Composio for managed OAuth (security guide)
~15% of community skills contain malicious instructions
- 341 malicious skills discovered in "ClawHavoc" campaign (Atomic Stealer malware)
- 280+ skills leak API keys and PII (Snyk)
- VirusTotal scanning now integrated
- Always audit skills before installation
- centminmod/explain-openclaw - Deep security analysis (73 files, multi-AI)
- Cisco - AI Agents Are a Security Nightmare
- Bitdefender - Malicious Skill Trap
- Snyk - 280+ Leaky Skills
- The Register - Security Issues
- SecurityWeek - Hijack OpenClaw AI Assistant
- SoC Radar - CVE-2026-25253 Analysis
- Runzero - OpenClaw RCE Vulnerability
- Depthfirst - 1-Click RCE to Steal Data and Keys
- VirusTotal - From Automation to Infection - 7.1% of ClawHub skills exposed credentials
- Cyera - The OpenClaw Security Saga
- Vectra - When Automation Becomes a Digital Backdoor
- Belgium CCB Advisory
- University of Toronto Advisory
| Provider | Config Key | Free Tier? | Notes |
|---|---|---|---|
| Anthropic | ANTHROPIC_API_KEY |
No | Primary, best quality |
| OpenAI | OPENAI_API_KEY |
No | GPT-4, GPT-4o |
GOOGLE_API_KEY |
Yes (Gemini Flash-Lite) | Cheapest hosted | |
| OpenRouter | OPENROUTER_API_KEY |
No | 200+ models unified API |
| DeepSeek | DEEPSEEK_API_KEY |
No | Excellent value |
| Moonshot | MOONSHOT_API_KEY |
No | Kimi K2.5 (great agentic) |
| MiniMax | MINIMAX_API_KEY |
No | M2.1 native support |
| Fireworks AI | Built-in | No | Model hosting |
| Workers AI | Built-in | Yes (limits) | Cloudflare models |
| Ollama | Local | Free | Requires local hardware |
| LM Studio | Local | Free | Requires local hardware |
| vLLM | Local | Free | High-performance inference |
| Channel | Setup | Auth Method | Notes |
|---|---|---|---|
| Telegram | 2 min | BotFather token | Fastest setup, supports Topics for parallel threads |
| 2 min | QR code scan | Separate phone number recommended | |
| Discord | 5 min | Bot application | Developer portal required |
| Slack | 5 min | Bot token (Socket or HTTP mode) | api.slack.com/apps |
| Signal | 5 min | Direct integration | Privacy-focused |
| iMessage | 10 min | BlueBubbles bridge | macOS only |
| Microsoft Teams | 5 min | Enterprise integration | Business users |
| Google Chat | 5 min | Workspace integration | Google Workspace |
| Matrix | 5 min | E2E encryption | Best for privacy |
| Zalo | 5 min | Direct integration | Popular in Vietnam |
| WebChat | Built-in | Gateway token | Browser-based interface |
curl -fsSL https://ollama.com/install.sh | sh
ollama pull llama3.1{ "provider": "openai", "baseUrl": "http://localhost:11434/v1", "model": "llama3.1" }{ "provider": "openai", "baseUrl": "http://localhost:1234/v1", "model": "your-model-name" }{ "provider": "openai", "baseUrl": "http://localhost:8000/v1", "model": "your-model-name" }| Plugin | Source | Features |
|---|---|---|
| openclaw-mcp-plugin | GitHub | HTTP/SSE transport, multi-server, unified interface |
| openclaw-mcp-adapter | npm | Registers MCP tools as native agent tools |
| Feature | Description |
|---|---|
| OpenClaw Voice | Self-hosted browser-based voice chat (Whisper STT + ElevenLabs TTS) |
| Android App | Customizable wake words, long-press home button activation |
| Voice Call Plugin | Telephony via Twilio, Telnyx, Plivo |
| Always-On Speech | macOS/iOS/Android with ElevenLabs |
-
Home Assistant custom add-on with ha-mcp integration
-
Control lights, thermostats, IoT devices via messaging
-
Adjust boilers based on weather forecasts
-
Voice commands through WhatsApp/Telegram
-
Gmail: Real-time processing via Pub/Sub, smart filtering, auto-responses
-
Google Calendar: Event creation, viewing, sync, daily briefings
-
Apple Calendar: Via khal/vdirsyncer integration
-
Outlook: Calendar sync and management
Full suite access: Gmail, Calendar, Drive, Docs, Sheets. All data stays local — never passes through third-party services.
Webhooks: POST to /hooks/wake or /hooks/agent with Bearer token authentication. Hookdeck for secure tunnels with retries.
Cron Jobs: Interval-based (every 30 minutes) or Unix cron expressions (0 9 * * 1). Stored at ~/.openclaw/cron/jobs.json. Exponential retry backoff.
Heartbeat: Agent wakes every 30 min (configurable), reads HEARTBEAT.md, decides if action needed. Different from cron — "check these things periodically" vs "do this specific thing at this time."
A2UI (Agent-to-UI) visual workspace — agent can render real-time UI, charts, and diagrams. Embedded in macOS menu bar app using WKWebView. Borderless, resizable, auto-reloads on file changes.
Configure multiple agents with separate workspaces, personas, auth profiles, and sessions (no cross-talk). Route by channel, phone number, or personality.
| App | Platform | Status | Features |
|---|---|---|---|
| Menu Bar App | macOS | Available | Pixel lobster icon, voice wake, Canvas panel, TCC permissions |
| Crabwalk | Cross-platform | Available | Open-source companion app (crabwalk.app) |
| iOS/Android | Mobile | Beta | Camera, screen recording, notifications |
| Tool | Type | Features |
|---|---|---|
| openclaw dashboard | Built-in | Gateway info, stats, web UI |
| ClawController | Third-party | Real-time monitoring, task orchestration, agent chat |
| claw-dash | Community | Sessions, tokens, costs, CPU/RAM/disk metrics |
| Mission Control | Community | Convex + React, live logs, task tracking |
openclaw config backup
openclaw config restore config-2026-02-01-1600.yaml
# Manual backup
tar -czvf ~/openclaw_backup_$(date +%Y%m%d).tar.gz -C "$HOME" .openclawNever commit
~/.openclaw/to Git (contains secrets).
| Hardware | Tokens/sec | Notes |
|---|---|---|
| RTX 4090 | ~80 t/s | Best GPU performance |
| AMD MI300X | High | 192 GB memory, free credits available |
| Apple M2 Pro 64GB | Good | Large memory pool for bigger models |
| MacBook Air M2 | ~3.2 t/s | CPU-only, usable |
| Typical CPU | ~13.5 t/s | Non-time-critical tasks |
| Oracle ARM (7B) | 5-10 t/s | Free tier, acceptable |
| Raspberry Pi 5 | 2-5 t/s | Slow but works |
| Model Size | RAM Required | GPU VRAM |
|---|---|---|
| 3B (quantized) | 4 GB | 4 GB |
| 7B (quantized) | 8 GB | 8 GB |
| 13B (quantized) | 16 GB | 12-16 GB |
| 70B (quantized) | 64 GB | 48 GB+ |
Bottleneck: Memory bandwidth and VRAM capacity, not CPU speed. Sweet spot: 7B models.
Complementary tools, not competitors. Many engineers run both simultaneously.
| Feature | Claude Code | OpenClaw |
|---|---|---|
| Lives in | Terminal / IDE | Messaging apps |
| Focus | Coding, development | Automation, integration |
| Memory | Session-based | Persistent (24/7) |
| Operation | Interactive | Autonomous |
| Runtime | Per-session | Always-on daemon |
| Integration | Git, tests, dev tools | Email, calendar, messaging, home automation |
| Best for | Multi-file refactoring, tests | Inbox monitoring, reminders, web scraping |
| Alternative | Type | Best For | Key Advantage |
|---|---|---|---|
| Nanobot | Lightweight (4K lines) | Minimalists | 99% smaller codebase |
| NanoClaw | Security-first | Security-conscious | Isolated Apple containers |
| memU | Memory-focused | Budget users | Local knowledge graph |
| Jan.ai | Offline chat | Privacy absolutists | 100% offline |
| AnythingLLM | Doc-to-chatbot | Knowledge bases | Turn PDFs into chatbots |
| Claude Code | Coding agent | Developers | Best coding assistant |
| Microsoft Copilot | Enterprise AI | Enterprise | Compliance, security |
| eesel AI | Business agent | Help desks | Customer service |
| Emergent | Personal AI | WhatsApp/Telegram | Simpler OpenClaw alternative |
| memU bot | Memory-first agent | 24/7 proactive assistant | Learns habits, acts without commands |
| Knolli.ai | Safe agentic AI | Security-first | Best OpenClaw alternative |
| Dify | LLM app platform | Agents, RAG, MLOps | Best debugging, workflow visualization |
| Langflow | RAG pipelines | Vector stores, MongoDB | MIT licensed, Astra DB native |
| Flowise | Chatbot builder | Rapid development | Prebuilt conversational templates |
| n8n | Workflow automation | Business processes | Visual workflow builder |
| LangBot | Multi-platform bots | IM integration | GitHub |
Created by OpenClaw agent "Clawd Clawderberg" (built by Matt Schlicht, Cofounder of Octane AI).
| Stat | Value |
|---|---|
| Launched | January 28, 2026 |
| Registered AI agents | 1.6M+ |
| Posts & responses | 7.5M+ AI-generated |
| Format | Reddit-style forum for AI agents |
| Human access | Read-only observation |
"Most interesting place on the internet right now" — Fortune
- URL: clawhub.ai | Skills: 5,705+ | GitHub: openclaw/clawhub
- URL: VoltAgent/awesome-openclaw-skills | Skills: 3,009 curated
Lobster — typed workflow runtime with resumable approvals. Turns skills/tools into composable pipelines.
| Skill | Description | Source |
|---|---|---|
| Mixpost | Social media management integration | ClawHub |
| AIsa Skills | Streamlined AI agent deployment | PheMex |
| Luma Events | Search events, RSVP, sync to Google Calendar (no API needed) | X/@bilbeny |
| QMD Skill | Cuts token usage by 95% | X/@milesdeutscher |
| Supermemory | Unlimited memory for OpenClaw | ClawHub |
| Claude Team | Spawns visible terminal sessions instead of background | X/@jlehman_ |
| ClawRouter | Smart LLM router — save 78% on inference costs, 30+ models | GitHub |
| Platform | Integration | Description |
|---|---|---|
| LangBot | Multi-platform | Agentic IM bots with OpenClaw, Dify, n8n, Langflow, Coze |
| OpenRouter | LLM routing | Access 30+ models with smart routing (openrouter/<author>/<slug>) |
openclaw skill install <skill-name>Always audit skills before installation. ~15% contain malicious instructions. Use VirusTotal-scanned ClawHub skills only.
| Mode | Description | Use Case |
|---|---|---|
| Extension Relay | Control existing Chrome tabs with logged-in sessions | Personal browsing |
| OpenClaw-managed | Isolated automation in dedicated browser | Web scraping, testing |
| Remote CDP | Distributed cloud deployments | Scale at cloud level |
Capabilities: CDP, ARIA snapshots, screenshots, tab management, click/type/drag, PDF export, video recording.
- Morning brief: Weather, calendar, headlines before checking your phone
- Inbox management: Process thousands of emails autonomously, unsubscribe from spam
- Grocery list: Auto-add items from household texts to shared documents
- Review PRs from your phone, run tests remotely, merge code
- Multi-instance Claude Code supervised by OpenClaw via Telegram + Tailscale
- Run coding agents while sleeping
- Control Philips Hue, Elgato, Home Assistant via messaging
- Adjust heating based on weather forecasts
- Security camera monitoring with alerts
- Calculate position sizes, manage stop-loss rules
- Trade alert logging and automation
- Voice notes to clean journal entries
- Platform-specific content: X threads, LinkedIn posts, blog articles
- Weekly meal planning in Notion (saves 1 hour/week)
-
Monitor school WhatsApp groups, filter noise
-
Run face recognition on photos, send daily digest to parents
- Official Getting Started
- Codecademy Tutorial (20 min)
- freeCodeCamp Full Tutorial
- Master OpenClaw in 30 Minutes
- How to Actually Tame It
- Beginner's Guide (5 min)
- 40 Tips & Tricks
- DataCamp - Control Your PC from WhatsApp
- DEV - Ultimate Guide for Developers 2026
- Macaron - Clean Setup + Verification
- BitLaunch - Install on macOS, Linux, Windows
- StackViv - Complete 2026 Guide
- Habr - Don't Launch Before Reading This
- AI/ML API - Secure Local Docker Setup
- AImaker - Is OpenClaw Worth the Hype? 10 Days
- AI/ML API - Real-World Use on $5 VPS
- DEV - I Tried the 'Free' AI Agent — $500 Reality Check
- DigitalOcean | AWS | Pulumi + Tailscale
- GCP | Azure | Oracle Free Tier
- Hostinger | Alibaba Cloud
- Vultr | Milvus/Slack
- Slack Integration | Notion Integration
- GitHub PR Automation
- Tailscale VPN Setup
- Custom Skill Creation
- OpenRouter Integration | DeepSeek via OpenRouter
- 50+ Official Extensions Guide
- Secure on Cloudflare | Proflead Guide
- Security & Cost Control | Composio Controls
- 7 Security Best Practices
- @ihtesham2005 - Setup Guide That Should Have Existed
- @VittoStack - Security-First Setup on Raspberry Pi via Tailscale
- @JordanLyall - The Security-First Guide I Wish Existed
- @harshil1712 - How I Setup OpenClaw on Raspberry Pi
- @BillDA - My Safe, Sandboxed Setup for OpenClaw
- Architecture: OpenClaw vs Claude Code
- Memory Architecture Explained
- How Memory System Works
- centminmod/explain-openclaw - Multi-AI security analysis (73 files)
- Cron Jobs Automation Guide
| Title | Platform | Duration | Topics |
|---|---|---|---|
| OpenClaw Full Tutorial for Beginners | freeCodeCamp (YouTube) | Long | Installation, models, memory, Docker, skills |
| Master OpenClaw in 30 Minutes | Creator Economy (YouTube) | 30 min | Safe setup, 5 real use cases, memory |
| How OpenClaw's Creator Uses AI | Creator Economy | 1 hour | Full demo with Peter Steinberger |
| OpenClaw Google Workspace Setup | YouTube | Medium | Gmail, Calendar, Drive integration |
| Udemy - OpenClaw on Azure Linux VM | Udemy | Medium | Azure deployment |
| Udemy - OpenClaw Complete (Japanese) | Udemy | Long | LINE + Docker, comprehensive |
| Platform | URL | Cost | Description |
|---|---|---|---|
| OpenClaw Academy | openclaw.academy | Free | All courses free, no signup, interactive quizzes |
| OpenClaw Academy Viewer | learn.openclaw.academy | Free | Interactive course viewer |
| freeCodeCamp | YouTube | Free | Comprehensive 1-hour course |
| Codecademy | Tutorial | Free | 20-minute install guide |
| GitHub Course | kiankyars/openclawcourse | Free | 1-hour crash course |
| Udemy | Azure VM Course | Paid | Azure Linux VM deployment |
| Platform | URL | Members | Best For |
|---|---|---|---|
| Discord | OpenClaw Discord | 8,000+ | Quick questions, troubleshooting |
| Discourse (CoClaw) | coclaw.com | 15,000+ | Long-form discussion, bug reports |
| X/Twitter Community | OpenClaw Community | 12,200+ | News, updates, demos |
| GitHub Issues | openclaw/openclaw/issues | N/A | Bug reports, feature requests |
| GitHub Discussions | openclaw/openclaw/discussions | N/A | Q&A, show-and-tell |
| Hacker News | Search | N/A | Technical discussion |
- CNBC - From Clawdbot to Moltbot to OpenClaw
- TechCrunch - AI assistants building their own social network
- Fortune - 'Most interesting place on the internet'
- VentureBeat - What the OpenClaw moment means for enterprises
- Nature - OpenClaw AI chatbots are running amok
- The Register - Clouds rush to deliver OpenClaw
- The Register - Security issues | Skills marketplace leak
- Hacker News - One-Click RCE | VirusTotal scanning
- Cisco - Security Nightmare
- Bitdefender - Malicious Skills
- Snyk - 280+ Leaky Skills
- Rest of World - Moltbot Finds Fans in China and Silicon Valley
- Trending Topics EU - Weekend Project Became Open-Source Sensation
- SlashGear - The Hottest New AI Agent Is Here
- Wikipedia - OpenClaw
- Wikipedia - Moltbook
- The Conversation - Why DIY AI Agent Feels New But Isn't
- XDA - Please stop using OpenClaw
- Pragmatic Engineer - "I ship code I don't read"
- Scientific American - AI Agent That Runs Your Computer
- Forkable - Clawdbot, Moltbot, OpenClaw... oh my!
- HN - OpenClaw Renamed Again
- DEV - From Moltbot to OpenClaw
| Issue | Cause | Fix |
|---|---|---|
| Crash during onboarding | < 2 GB RAM | Upgrade to 4 GB+ RAM |
| API key not working | Incorrect format or expired | Re-enter via openclaw auth set |
| Telegram not connecting | Using Cloudflare Workers | Switch to VPS (needs persistent connection) |
| Slow inference (3.5s/token) | CPU-only with large model | Use smaller model or add GPU |
| Port conflict on 18789 | Another service using port | openclaw gateway --port 18790 |
| Docker sandbox issues | Permissions or config | Run openclaw doctor |
| Skills not loading | Corrupt installation | Reinstall skill |
| Messages lost between channels | Known bug | Update to latest version |
| High token consumption | ~35,600 tokens per message overhead | Reduce workspace files |
| Rate limiting (#5159) | Aggressive retry loops | Update (exponential backoff fix) |
openclaw doctor # Common issues
openclaw security audit --deep # Deep security scan
openclaw security audit --fix # Auto-fix safe issues
node --version # Must be 22+- OpenClaw Cost Calculator
- API Usage and Costs Docs
- eesel.ai Pricing Guide
- Deploy Cost Guide: $0-8/month
| Budget | Strategy |
|---|---|
| $0 | Oracle Cloud + Gemini free tier + Ollama |
| $6 | Agent37 ($1) + Claude Haiku ($5) |
| $19 | Hetzner ($4) + mixed models ($15) |
| $50 | DigitalOcean ($12) + Claude Sonnet ($38) |
| $100 | Managed hosting ($24) + heavy API ($76) |
| Contributor | Role | GitHub |
|---|---|---|
| Peter Steinberger | Creator | @steipete |
| Mario Zechner | Pi creator, security pen-tester | @badlogicc |
| Maxim Vovshin | Blogwatcher skill | @Hyaxia |
| Nacho Iacovino | Location parsing | @nachoiacovino |
| 500+ contributors | Community | All contributors |
| Repository | Description |
|---|---|
| openclaw/openclaw | Main repository (176K+ stars) |
| openclaw/clawhub | Official skill directory |
| openclaw/lobster | Typed workflow runtime |
| cloudflare/moltworker | Cloudflare Workers adaptation |
| VoltAgent/awesome-openclaw-skills | 3,009 curated skills |
| centminmod/explain-openclaw | Deep security analysis (73 files) |
| lunarpulse/openclaw-mcp-plugin | MCP server integration |
| essamamdani/openclaw-coolify | Coolify PaaS template |
| phioranex/openclaw-docker | Docker deployment configs |
| gavrielc/nanoclaw | Security-first alternative |
| BlockRunAI/ClawRouter | Smart LLM router (save 78% on costs) |
| langbot-app/LangBot | Multi-platform IM bot with OpenClaw |
| kiankyars/openclawcourse | 1-hour crash course |
| Chart | Source | Features |
|---|---|---|
| serhanekicii/openclaw-helm | GitHub | Built on bjw-s app-template |
| waTeim/openclaw-helm | GitHub | Multi-channel support |
| khal3d/openclaw | GitHub | Docker & HELM deployment |
"It is not enterprise software. There is no promise of quality, no vendor support, no SLA… it ships without authentication enforced by default." — Gartner
- Vectra - When Automation Becomes a Digital Backdoor
- Palo Alto Networks — OpenClaw presents a "lethal trifecta": access to private data + exposure to untrusted content + ability to communicate externally while retaining memory
- Cyera - How AI Adoption Outpaced Security Boundaries
For organizations needing compliance, SLAs, and vendor support, consider:
- Kilo Claw — SSO, audit logs, pay-as-you-go
- ClawCloud Business — $129/mo, dedicated resources
- OpenClaw Hosting Business — Team/Business tiers
- Self-hosted with NanoClaw (security-first fork) in isolated containers
OpenClaw has seen explosive adoption in China with support from major tech companies.
| Company | Integration | Details |
|---|---|---|
| Alibaba Cloud | Simple Application Server | 19 regions, from $4/mo |
| Tencent Cloud | Lighthouse Service | One-click install |
| ByteDance | Volcano Engine | Cloud hosting support |
- Native Feishu (飞书) and Lark support — first official Chinese chat client integration
- EvolutionAI Hub - OpenClaw v2026.2.2
- Rest of World - Moltbot Finds Fans in China and Silicon Valley
- China's Tech Giants Opening Doors to OpenClaw
| Version | Date | Key Changes |
|---|---|---|
| v2026.2.2 | Feb 2026 | Feishu/Lark support, onchain integrations |
| v2026.2.1 | Feb 2026 | Security hardening, streaming stability, path traversal fixes |
| v2026.1.30 | Jan 30, 2026 | CVE-2026-25593 & CVE-2026-25475 patches |
| v2026.1.29 | Jan 29, 2026 | CVE-2026-24763, CVE-2026-25253 patches, VirusTotal scanning |
| v2026.1.24-1 | Jan 24, 2026 | DigitalOcean 1-Click base version |
Contributions welcome! Please read the contribution guidelines first.
- Add a resource via pull request
- Report broken links via issues
- Suggest new sections or improvements
This list is released under CC0 1.0 Universal (Public Domain).
