🔥 2M+ visitors in a week | The AI agent taking the world by storm

Secure Your OpenClaw in 5 Minutes

Complete hardening guide + security audit checklist for AI agents. Protect against prompt injection, configure gateways, and deploy with confidence.

5 Core Guides
53 Checklist Items
5 min Quick Audit
🛡️
Security
Hardened
🦞
New Name, Same Lobster! Moltbot is now OpenClaw. Read the announcement or migrate now.

Quick Security Wins

High-impact improvements you can make right now

Run the Security Audit

openclaw security audit

Restrict File Permissions

chmod 600 ~/.openclaw/*

Use Opus 4.5

Best prompt injection resistance

Enable DM Allowlists

Control who can interact

Latest Security News

Stay updated with open source AI security

Frequently Asked Questions

How do I install OpenClaw?

Install OpenClaw using npm: npm install -g openclaw@latest. Requires Node.js 18 or higher. Then run 'openclaw onboard --install-daemon' to set up authentication, gateway configuration, and optional background service (systemd on Linux, launchd on macOS).

What is the OpenClaw security audit command?

Run 'openclaw security audit' to check your OpenClaw configuration for security issues. This command verifies file permissions, authentication settings, gateway configuration, and plugin security. Run it before production deployment and after any configuration changes.

How often should I run the OpenClaw security audit?

Run security audits weekly for development environments, before production deployments, after any configuration changes, when adding new team members, and after migrating from Moltbot/Clawdbot. Use the Security Audit Checklist for comprehensive manual review.

What does the OpenClaw security audit check?

The security audit checks: file permissions (should be 600 for configs), gateway binding (should be 127.0.0.1, not 0.0.0.0), authentication configuration (token or password mode), DM policies and allowlists, symlinks in config directories, and legacy Moltbot installations.

What are the main security areas covered by OpenClaw?

OpenClaw security covers: inbound access controls (DM policies, group allowlists), tool blast radius (sandbox settings, elevated permissions), network exposure (gateway binding, authentication), browser controls (download directories, URL restrictions), disk hygiene (file permissions, symlinks), and model selection for prompt injection resistance.

What is prompt injection in OpenClaw?

Prompt injection is an attack where malicious input tricks your OpenClaw agent into performing unintended actions. Protect against it by using Claude's Opus 4.5 (best resistance), enabling DM allowlists with denyByDefault: true, running regular security audits, restricting elevated tools, and using secure onboarding via 'openclaw onboard'.

How do I access OpenClaw documentation?

Visit securemolt.com for comprehensive OpenClaw security guides covering installation, gateway hardening, security audits, prompt injection defense, and migration from Moltbot/Clawdbot. For quick setup, run 'openclaw onboard --install-daemon' to start the interactive setup wizard.

Can OpenClaw security audit be automated?

Yes, 'openclaw security audit' can be integrated into CI/CD pipelines (GitHub Actions, GitLab CI), scheduled maintenance tasks, or automated health check routines. See the Migration Guide for pipeline examples using 'npm install -g openclaw' followed by 'openclaw security audit'.

How do I migrate from Moltbot to OpenClaw?

Run 'npm install -g openclaw@latest' then 'openclaw migrate --from-moltbot'. This migrates your config from ~/.config/moltbot/ to ~/.openclaw/. You MUST configure authentication (auth: none is removed). Finally run 'openclaw onboard --install-daemon' and 'openclaw security audit' to verify.

What authentication modes does OpenClaw support?

OpenClaw requires authentication - auth: none has been permanently removed. Use token authentication (recommended, 256-bit random tokens with 'openssl rand -hex 32') or password authentication (Argon2 hashed via 'openclaw auth hash-password'). Store tokens in environment variables, never in code.

Ready to Deploy Securely?

For production deployments, we recommend a dedicated VPS with proper isolation.

Get Started with Digital Ocean →

Affiliate link - we may earn a commission at no extra cost to you