Securing OpenClaw Against ClawHavoc Supply-Chain Attacks

OpenClaw, previously known as Clawdbot and Moltbot, is a platform that allows autonomous AI agents to operate with direct access to file systems and terminals. This architecture significantly increases its attack surface, making it vulnerable to various threats. In February 2026, researchers discovered the ClawHavoc supply-chain attack, which exploited these vulnerabilities by embedding malicious skills within the ClawHub ecosystem. Approximately 12% of these skills were found to be harmful, disguised as useful tools while actually stealing digital identities.

The ClawHavoc attack operates through a structured kill chain. It begins with attackers uploading a malicious skill to ClawHub, specifically targeting the SKILL.md file. This file contains instructions that trick the agent into executing harmful commands, leading to the download of malware such as Atomic Stealer (AMOS) or keyloggers, which can compromise sensitive information like API keys and crypto wallets.

Aryaka AI>Secure provides a multi-layered defense against the ClawHavoc threat. It acts as a deep-packet, AI-aware MITM proxy that intercepts traffic throughout the attack chain. The first method of protection involves blocking the download of malicious skills by parsing the SKILL.md content and identifying harmful instructions before they reach the agent.

Additionally, Aryaka AI>Secure employs response semantic filtering to detect and block harmful commands generated by the LLM, preventing users from executing dangerous scripts. This second layer of defense ensures that even if a malicious skill is present, the AI’s output is scrutinized for potential manipulation.

Furthermore, Aryaka AI>Secure’s Secure Web Gateway (SWG) functionality proactively filters URLs, preventing users from accessing known malicious domains that host malware payloads. This real-time URL intelligence helps to stop threats before they can infiltrate the system.

Finally, if malware does execute, Aryaka AI>Secure acts as a gatekeeper, monitoring data exfiltration attempts. It scans outbound data for sensitive information and terminates sessions if it detects any anomalies, alerting the security team to potential breaches.

The ClawHavoc incident highlights the need for robust security measures in autonomous systems. By utilizing solutions like Aryaka AI>Secure, organizations can protect their OpenClaw agents from becoming entry points for attackers, ensuring they remain effective tools for productivity.