A comprehensive security toolkit designed to harden AI agent gateways and protect sensitive API keys and conversation data.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install moltbot-security
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install moltbot-security using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Moltbot Security Hardening?
Moltbot Security is an essential hardening framework for developers using AI agents like Moltbot, OpenClaw, and Cursor. Because these gateways are often designed for local use, exposing them to the internet without proper configuration can lead to the leak of private messages, LLM API keys, and full system shell access. By utilizing these Openclaw Skills, developers can implement a zero-trust architecture that ensures their vibe-coding setups remain private and resilient against external threats.
This skill addresses real-world vulnerabilities identified in research that found thousands of exposed gateways online. It provides both automated auditing tools and manual configuration paths to secure every layer of the agent's environment, from the network interface to file system permissions.
Moltbot Security Hardening Use Cases
- Securing private conversation histories from messaging apps like Telegram, WhatsApp, and iMessage.
- Protecting sensitive LLM credentials for Claude, OpenAI, and other providers.
- Preventing remote shell access and prompt injection attacks on local host machines.
- Enabling secure remote access to AI agents without opening dangerous public ports.
How Moltbot Security Hardening Works
- The security audit command scans the local environment for common misconfigurations like public port exposure and weak file permissions.
- The gateway is configured to bind strictly to loopback or LAN interfaces to prevent direct internet reachability.
- A robust authentication layer is established using 64-character hex tokens or secure passwords.
- File system permissions are restricted using chmod commands to ensure only the owner can read sensitive credential files.
- The skill integrates with Tailscale to create an encrypted tunnel for remote access, maintaining security while providing flexibility for Openclaw Skills.
Moltbot Security Hardening Setup
Install the security hardening skill via the hub:
clawdhub install NextFrontierBuilds/moltbot-security
Run a deep security audit to check your current posture:
openclaw security audit --deep
To automatically fix identified security issues, use the fix flag:
openclaw security audit --deep --fix
Moltbot Security Hardening Data Schema & Taxonomy
The skill manages security settings primarily within the Openclaw configuration directory. It ensures the following metadata and file taxonomy is followed:
| File/Directory | Purpose | Recommended Permission |
|---|---|---|
~/.openclaw/ |
Root configuration folder | 700 (Owner Read/Write/Exec) |
openclaw.json |
Gateway and authentication settings | 600 (Owner Read/Write) |
credentials/ |
Stored API keys and bot tokens | 700 (Owner Read/Write/Exec) |
Key configuration parameters include gateway.bind set to loopback and gateway.auth.mode set to token.
Moltbot Security Hardening Advanced Features
- Deep vulnerability auditing with the --deep flag to identify edge-case security risks.
- Support for Tailscale Serve mode to enable secure, zero-config remote access.
- Environment variable overrides for sensitive tokens using CLAWDBOT_GATEWAY_TOKEN.
- Network silence mode to disable mDNS and Bonjour broadcasting via CLAWDBOT_DISABLE_BONJOUR.
- Automated UFW firewall rule generation to block unauthorized traffic to port 18789 while allowing SSH and Tailscale.
SKILL.md
Loading
Related Openclaw Skills
MoltBot Best Practices
A definitive guide and configuration set for AI agents to ensure human-in-the-loop validation, fail-fast protocols, and reliable execution.
Crypto Fear & Greed Index Widget
A drop-in, real-time sentiment gauge for crypto dashboards and trading applications.
AI PDF Builder
A powerful AI-driven generator for creating YC-style legal documents, technical whitepapers, and professional reports from simple prompts.
Calibre Workspace Manager
A routing-optimized entry point for AI agents to interact with Calibre libraries for catalog reading and metadata management.
Package SEO Skill
A specialized toolkit for optimizing npm packages, GitHub repositories, and AI agent skills to maximize search engine and agent discoverability.
Prism Alerts
A real-time Solana token monitoring tool for tracking Pump.fun launches and graduations within the Openclaw Skills ecosystem.
METADATA
Featured*
