Five malicious skills removed from OpenClaw ClawHub marketplace
Security researchers at Palo Alto Networks' Unit 42 identified five malicious skills on OpenClaw's ClawHub marketplace that could steal credentials, evade detection, and conduct unauthorized financial transactions. OpenClaw removed the skills and banned the associated accounts after the researchers reported them.
What happened
Security researchers at Palo Alto Networks' Unit 42 identified five malicious skills on ClawHub, OpenClaw's dedicated marketplace, between February and May 2026. The skills were designed to appear legitimate while carrying malicious functionality.
The five skills fell into three threat categories: two delivered macOS infostealers that connected to command-and-control infrastructure; one skill had an inflated file size to exceed scanner thresholds and bypass ClawScan and VirusTotal detection; and two skills represented novel agentic threats including runtime agentic affiliate injection and agentic front-running used for financial gain.
Unit 42 reported all five skills to ClawHub for takedown. OpenClaw banned the associated accounts and deleted all of the skills.
Reported impact
Unit 42 researchers stated that if the malicious skills had spread across users of OpenClaw, they could have threatened the platform in use across scores of organizations.
The infostealers could steal credentials and access local files, APIs, and other resources on host systems, while the agentic threats could conduct unauthorized financial transactions.
Index analysis
The incident demonstrates that AI agent skill marketplaces have emerged as a significant supply chain attack surface, similar to traditional software supply chain vulnerabilities but novel to the agentic AI ecosystem.
The skills evaded ClawHub's security screening mechanisms including VirusTotal and ClawScan integration that were implemented in February 2026 following earlier malicious campaigns.
Governance controls
Following the February 2026 discovery of early malicious campaigns, ClawHub integrated VirusTotal and ClawScan for proactive screening of published skills and code-level analysis.
OpenClaw is collaborating with NVIDIA to provide documentation of what each skill does and to run NVIDIA's analysis tool on all skills.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.