More Malicious OpenClaw Skills Threaten AI Supply Chain

Unit 42 identified five malicious OpenClaw marketplace skills on ClawHub that perform credential theft (macOS infostealers), evade automated scanners by abusing file-size and README padding, and enable novel agentic financial abuse (affiliate injection and a meme-token pump-and-dump). The skills bypassed ClawHub defenses, were removed after reporting, and researchers warn that such agentic-supply-chain threats can scale across organizations using OpenClaw unless stronger provenance, runtime monitoring, and line-by-line audits are enforced.