OpenClaw has integrated VirusTotal’s malware scanning into its ClawHub marketplace to address security concerns that arose after the discovery of 341 malicious skills and unauthorized deployments. The platform, which has gained over 150,000 GitHub stars since its launch in November 2025, allows developers to create skills that run with full access to the AI agent’s tools. Security experts have flagged OpenClaw as “insecure by default,” and the new integration aims to automatically scan all published skills for malware before they go live. Skills deemed “benign” are approved, while suspicious ones receive warnings, and malicious skills are blocked, with daily re-scans for active skills. The integration relies on VirusTotal’s Code Insight, which analyzes skill packages for malicious behavior beyond mere virus signatures. It assesses whether skills attempt to download external code, access sensitive data, or perform unsafe actions. Despite this advancement, OpenClaw acknowledges the limitations of malware scanning. Vulnerabilities like prompt injection—which can hijack agent behavior without traditional malware signatures—remain significant risks. Security professionals warn that threats like logic abuse and misuse of legitimate tools are not addressed by the current scanning process. The urgency for this initiative follows a Koi Security audit that revealed the extent of the problem, with a campaign dubbed “ClawHavoc” exposing numerous malicious tools. A Cornell University report found that 26% of skills contained vulnerabilities. Security vendor Noma reported that over half of its enterprise clients had given OpenClaw privileged access within a single weekend, raising alarms about shadow IT deployments. While the VirusTotal integration is a necessary step toward improving marketplace safety, experts emphasize that it’s just the beginning of a broader security strategy that needs to be developed further.