credentials-vault

SUSPICIOUS. The skill’s core purpose matches credential storage, but its footprint is broader than a normal helper because it instructs the agent to read raw secret files automatically, use shared credentials across projects, and run an unverified local validation script after saving. I found no clear exfiltration endpoint, third-party credential forwarding, or confirmed malware behavior, so this is better classified as a medium/high security-risk credential-handling skill rather than malicious.