OpenClaw & Moltbook FAQ

After installation, check these: 1. Run `openclaw health` or `openclaw status` (CLI) or check the health status in Desktop 2. Look for green checkmarks on all services 3. Verify the gateway is accessible (default: http://127.0.0.1:18789) 4. Check logs for any error messages If all checks pass, you're ready to configure channels.

Top 5 startup issues: 1. Port already in use (another app using 18789) 2. Missing or invalid API keys 3. Insufficient permissions (try running as admin) 4. Firewall blocking connections 5. Outdated dependencies (Node.js v22+ required) Check the logs first - they usually point to the exact issue.

Solutions: 1. Find and stop the conflicting process: • Mac/Linux: `lsof -i :18789` then `kill [PID]` • Windows (WSL): `lsof -i :18789` then `kill [PID]` 2. Or change OpenClaw's port: • Use `openclaw gateway --port 18790` when starting • Or specify `--gateway-port 18790` during onboard

Try these steps in order: 1. Check logs: `openclaw logs --follow` to see error messages 2. Reconfigure: run `openclaw configure` to reset settings 3. Reinstall: `npm install -g openclaw@latest` 4. Check release notes for breaking changes 5. If still failing, report with logs on GitHub Issues

OpenClaw primarily uses `~/.openclaw/openclaw.json` for configuration. Common issues: 1. Service not restarted after config changes 2. Wrong configuration file path 3. JSON syntax errors (missing commas, quotes) 4. Environment variables should be passed via onboard parameters Verify config: run `openclaw status` to see active settings.

Think of it like a restaurant: • Model = The chef (AI that generates responses) • Channel = The entrance (WhatsApp, Telegram, etc.) • Gateway = The host stand (receives and routes requests) • Skills = Special abilities (web search, code execution, etc.) You need at least one model and one channel. Gateway runs automatically. Skills are optional enhancements.

The easiest way is to run the interactive setup wizard: ``` $ openclaw onboard ``` The wizard will guide you through: • Selecting an AI model provider (Anthropic/OpenAI etc.) • Entering your API key • Installing and starting the gateway service After setup, connect a channel with `openclaw channels login telegram`. Default gateway binds to 127.0.0.1:18789 (local access only - safe).

In Desktop: Settings → Skills → Toggle on/off In CLI: • List available skills: `openclaw skills list` • Install a skill: `openclaw skills install [skill-name]` • Enable a skill: `openclaw skills enable [skill-name]` Safety tips: • Disable skills you don't need • Review skill permissions before enabling • Test new skills in a safe environment first • Some skills can execute code - enable with caution

Switching models: 1. Run `openclaw onboard` again to reconfigure with new provider 2. Or edit `~/.openclaw/openclaw.json` agents.defaults.model settings 3. Restart the service: `openclaw gateway status` to verify If new model fails: • Run `openclaw models status` to check model connection • Check API key validity on provider's website • Verify model name spelling (e.g., `anthropic/claude-opus-4-5`) • Check rate limits or quota on your account

Security checklist: ✓ Keep gateway on localhost (127.0.0.1) by default ✓ Never expose port directly to public internet ✓ If remote access needed, use: • VPN or SSH tunnel • Reverse proxy with authentication • Cloudflare Tunnel ✓ Enable authentication if available ✓ Use HTTPS in production

Try these fixes: 1. Make sure phone and computer are on same network 2. Refresh the QR code (they expire quickly) 3. Update WhatsApp to latest version on phone 4. Clear browser cache if using web interface 5. Try scanning in a well-lit environment 6. Disable any VPN temporarily If still failing, try the phone linking code method instead.

Session issues are usually caused by: 1. Phone going offline or sleeping 2. WhatsApp logged out on phone 3. Using WhatsApp Web on another device 4. Network instability Fixes: • Keep phone connected to stable WiFi • Disable battery optimization for WhatsApp • Don't open WhatsApp Web elsewhere • Check for WhatsApp app updates

Group behavior is usually intentional: • Check if group replies are enabled in config • Bot may require @mention to respond in groups • Default config may disable group messages for safety To enable: • Enable group replies in Desktop settings • Or manage group permissions via `openclaw pairing` • Test with a small group first to ensure it works

Telegram bots have Privacy Mode enabled by default: 1. Open @BotFather 2. Send /mybots 3. Select your bot 4. Bot Settings → Group Privacy → Turn OFF Alternatively, make the bot a group admin - admins can see all messages regardless of privacy mode.

Simple steps: 1. Open Telegram, search for @BotFather 2. Send /newbot 3. Choose a name (display name) 4. Choose a username (must end in 'bot') 5. Copy the API token provided 6. Add token to your OpenClaw config Optional: Use /setcommands to add command hints.

Usually an intents issue: 1. Go to Discord Developer Portal 2. Select your application → Bot 3. Enable these Privileged Intents: • MESSAGE CONTENT INTENT • SERVER MEMBERS INTENT (if needed) 4. Save and restart your bot Also check: bot has Read Messages permission in the channel.

Required intents: • Guilds - basic server info • Guild Messages - receive messages • Message Content - read message text (PRIVILEGED) Optional: • Guild Members - user info (PRIVILEGED) • Direct Messages - DM support Note: Privileged intents require verification for bots in 100+ servers.

Minimum Bot Token Scopes: • chat:write - send messages • app_mentions:read - respond to @mentions • channels:history - read channel messages • im:history - read DM messages Event Subscriptions: • message.channels • message.im • app_mention Start minimal, add more only when needed.

401 means authentication failed. Check: 1. API key is correct (no extra spaces) 2. Key hasn't expired or been revoked 3. Key has required permissions/scopes 4. You're not hitting rate limits 5. Account has sufficient credits/quota Test your key directly with provider's API first.

Diagnose step by step: 1. Test model directly (curl/Postman) - if slow, it's the model 2. Check your internet speed and latency 3. Try a different model (GPT-3.5 is faster than GPT-4) 4. Check provider status page for outages Fixes: • Increase timeout settings • Use streaming for long responses • Try a different model provider

Check logs for these patterns: • 'FATAL' or 'PANIC' - critical errors • 'OOM' - out of memory • 'ECONNREFUSED' - can't connect to dependency • Stack traces - code errors Common fixes: • Increase memory allocation • Fix configuration errors • Update to latest version • Check external service availability

Config troubleshooting: 1. Did you restart the gateway? (Required for most changes) 2. Is JSON syntax correct? (Check commas, quotes) 3. Right config file? (Should be `~/.openclaw/openclaw.json`) 4. Is gateway running? (`openclaw gateway status`) Verify config: • `openclaw status` - check overall status • `openclaw models status` - check model connections • `openclaw health` - health check

Log locations: • Desktop: View → Logs (or Settings → Open Logs Folder) • CLI: `~/.openclaw/logs/` or stdout • Docker: `docker logs [container]` Key error keywords: • ERROR, FATAL, PANIC - critical issues • timeout, ETIMEDOUT - connection issues • 401, 403, 429 - auth/rate limit issues • ENOENT - missing file/path

Include in your report: 1. OpenClaw version (`openclaw --version`) 2. OS and version 3. Steps to reproduce 4. Expected vs actual behavior 5. Relevant logs (remove sensitive info!) 6. Config snippet (remove API keys!) Submit at: GitHub Issues page For security issues: email security contact directly

Risks of public exposure: • Anyone can send requests to your AI • API keys could be extracted • Potential for abuse (spam, illegal content) • Cost explosion from unauthorized usage • Legal liability for bot's actions Safe remote access options: • VPN (recommended) • SSH tunnel • Cloudflare Tunnel • Reverse proxy with auth

Safety checklist: 1. Is it from official/verified source? 2. Check the source code if available 3. Read reviews/issues from other users 4. Understand what permissions it requests 5. Test in isolated environment first Red flags: • Requests unnecessary permissions • Obfuscated/minified code • No source code available • No community feedback

Act immediately: 1. REVOKE the key on provider's website NOW 2. Generate a new key 3. Update all your configs with new key 4. Check usage history for unauthorized calls 5. Review how it leaked (git history? public repo?) Prevention: • Never commit keys to git • Use .env files (gitignored) • Use secrets management in production

Secure defaults: • Gateway bound to localhost only (127.0.0.1) • Skills disabled by default • Group messages disabled initially • Rate limiting enabled • Logging enabled (for audit) • Minimal permissions for channel bots Remember: start restrictive, loosen only when needed.