virtuals-protocol-acp

SUSPICIOUS: the skill is coherent with Virtuals ACP usage, but it asks the agent to execute wallet-backed blockchain jobs using a raw private key and some autonomous retry/create behavior. No obvious malicious exfiltration is shown, yet the combination of high-value credentials, real-world on-chain effects, and mildly ambiguous execution provenance makes this a medium-high security risk rather than benign.