# SafeClaw

> SafeClaw is the security and observability platform for OpenClaw, the free, open-source, self-hosted AI assistant. SafeClaw provides real-time monitoring, vulnerability scanning, and security hardening for autonomous AI agents. It helps developers and teams safely deploy AI assistants by offering shell logging, dependency scanning, permission scoping, and prompt injection protection.

SafeClaw sits between your autonomous AI agents and the systems they interact with, giving you complete visibility and control. The platform addresses critical security risks including prompt injection attacks, malicious dependency installation, unauthorized file system access, and credential exposure.

OpenClaw is a free, open-source, self-hosted AI assistant created by Peter Steinberger. It supports multiple AI models (Claude, GPT, Ollama, DeepSeek, Kimi/Moonshot), integrates with messaging platforms (WhatsApp, Telegram, Discord, Signal, iMessage), and can be deployed on Docker, VPS, Mac Mini, or any Linux server.

## Core Features

- [Real-time Shell Logging](https://safeclaw.io/): Watch agents think and type in real-time, replay every session to debug logic errors, and maintain complete visibility into autonomous agent actions
- [Vulnerability Shield](https://safeclaw.io/): Flag and block unsafe npm/pip packages before agents install them, protecting against malicious dependencies
- [Permission Scoping](https://safeclaw.io/): Define exactly which folders, URLs, and APIs agents can access with sandbox mode enforcement
- [PromptGuard](https://safeclaw.io/): Input filtering to detect and block prompt injection attacks
- [OpenClaw Skills Security Scanner](https://safeclaw.io/openclaw-skills-security-scanner): Scan OpenClaw skills for vulnerabilities, malicious code, prompt injections, malicious commands, and data exfiltration attempts

## Setup Guides

- [Quick OpenClaw Setup](https://safeclaw.io/quick-openclaw-setup): Under-1-minute setup guide with step-by-step installation for OpenClaw with WhatsApp and Anthropic Claude
- [Secure OpenClaw Setup](https://safeclaw.io/secure-openclaw-setup): Enterprise-grade security hardening guide with risk assessments, severity levels, and detailed mitigation strategies

## Getting Started

- [What Is OpenClaw?](https://safeclaw.io/blog/what-is-openclaw): Complete guide to the self-hosted AI assistant
- [How to Install OpenClaw](https://safeclaw.io/blog/how-to-install-openclaw): Step-by-step installation guide for all platforms
- [How to Set Up OpenClaw](https://safeclaw.io/blog/openclaw-setup): Complete configuration guide from zero to running
- [System Requirements](https://safeclaw.io/blog/openclaw-system-requirements): Hardware and software requirements guide
- [How to Use OpenClaw](https://safeclaw.io/blog/how-to-use-openclaw): Complete beginner's guide

## Security

- [Is OpenClaw Secure?](https://safeclaw.io/blog/openclaw-security): Security best practices and hardening guide
- [Is OpenClaw Safe to Use?](https://safeclaw.io/blog/is-openclaw-safe): Privacy, security, and risk analysis

## AI Model Integration

- [Supported AI Models](https://safeclaw.io/blog/openclaw-models): Complete model compatibility guide
- [Ollama Integration](https://safeclaw.io/blog/ollama-openclaw): Run local AI models without cloud APIs
- [OpenAI GPT Integration](https://safeclaw.io/blog/openclaw-openai): GPT model setup and configuration
- [Anthropic Claude Integration](https://safeclaw.io/blog/openclaw-anthropic): Claude API setup and configuration
- [DeepSeek Integration](https://safeclaw.io/blog/openclaw-deepseek): DeepSeek model integration guide
- [Kimi AI Integration](https://safeclaw.io/blog/openclaw-kimi): Moonshot AI model integration guide

## Deployment

- [Docker Deployment](https://safeclaw.io/blog/openclaw-docker): Containerized deployment guide
- [VPS Deployment](https://safeclaw.io/blog/openclaw-vps): Cloud server hosting guide
- [Mac Mini Server](https://safeclaw.io/blog/openclaw-mac-mini): Always-on AI server setup

## Messaging Integrations

- [Telegram Integration](https://safeclaw.io/blog/openclaw-telegram): Messaging bot integration guide
- [Discord Integration](https://safeclaw.io/blog/openclaw-discord): Complete bot integration guide
- [Discord Developer Portal Setup](https://safeclaw.io/blog/openclaw-discord-developer-portal): Step-by-step developer portal guide

## Developer Tools

- [Claude Code Integration](https://safeclaw.io/blog/openclaw-claude-code): Anthropic coding integration guide
- [Codex Integration](https://safeclaw.io/blog/openclaw-codex): OpenAI Codex integration for coding
- [OpenClaw CLI](https://safeclaw.io/blog/openclaw-cli): Command line reference and examples
- [OpenClaw Dashboard](https://safeclaw.io/blog/openclaw-dashboard): Control UI features and overview
- [Browser Automation](https://safeclaw.io/blog/openclaw-browser): Web browsing and scraping guide

## Architecture

- [OpenClaw Gateway](https://safeclaw.io/blog/openclaw-gateway): Understanding the core architecture
- [OpenClaw Skills](https://safeclaw.io/blog/openclaw-skills): How to extend your AI assistant with plugins
- [MoltBot Framework](https://safeclaw.io/blog/moltbot): Understanding the OpenClaw bot framework
- [Best OpenClaw Skills](https://safeclaw.io/blog/awesome-openclaw-skills): The awesome skills collection for 2026

## Optional

- [GitHub Repository Guide](https://safeclaw.io/blog/github-openclaw): Repository guide and source code overview
- [Reddit Community](https://safeclaw.io/blog/openclaw-reddit): Community reviews and discussions
- [Peter Steinberger](https://safeclaw.io/blog/openclaw-peter-steinberger): The creator and visionary behind OpenClaw
- [Moonshot AI Integration](https://safeclaw.io/blog/openclaw-moonshot-ai): Kimi model integration guide
- [How to Uninstall OpenClaw](https://safeclaw.io/blog/how-to-uninstall-openclaw): Complete removal and cleanup guide