sealevel-guard-review

SUSPICIOUS. The skill is purpose-aligned as a Solana audit orchestrator, but it is still high risk because it fetches and analyzes untrusted external code and produces high-consequence trust recommendations for deployment and capital allocation. I see no clear credential theft or malicious exfiltration, so this is not malware, but it should be treated as a powerful security-review skill with meaningful indirect prompt-injection and autonomy risk.