OpenClaw security assessment
A Hacker News article published last week revealed that over 280,000 OpenClaw instances are exposed to serious security risks. According to the report, 12% of skills on ClawHub contain malware, and the platform has no built-in protection against prompt injection or data exfiltration.
How attackers are exploiting OpenClaw
Security researchers have identified four main attack vectors currently being used against OpenClaw users:
- Prompt injection — attackers craft prompts that trick agents into revealing API keys or executing harmful commands.
- Malicious skills — nearly one in eight skills on the official ClawHub contain backdoors, cryptominers, or data stealers.
- Data exfiltration — the link preview feature can leak sensitive data without the user even clicking.
- Fake installers — info-stealers disguised as OpenClaw installers are spreading via forums and social media.
A response from the community
In response to the Hacker News report, independent security researcher Pedro Jose (founder of Cyber Flex Consultant and @StackOfTruths) released OpenClaw Security Sentinel — a tool that adds real-time protection to OpenClaw instances.
“I’ve been in cybersecurity for 10 years,” Jose told us. “OpenClaw is powerful, but it was built with zero security in mind. I built Sentinel because someone had to.”
Uses 6 AI agents to detect and block prompt injection, scan skills for malware, and prevent data leaks. New in v2.1: PostgreSQL brain for long-term memory.
What this means for OpenClaw users
If you’re running OpenClaw — whether for personal use or in production — you’re currently exposed. The platform itself has no security layer. Until official patches arrive, third-party tools like Sentinel are the only line of defense.
Jose is also offering a 20% discount for X subscribers. The full security assessment and Sentinel details are available on his website.
Leave a Reply