Citation and grounding checks for AI output: dangling references, unsupported counts, overclaims.
Verification confirms publisher identity (repo ownership), not code safety. The security scan covers known CVEs and suspicious install scripts — it cannot prove the absence of malicious code.
Citation and grounding checks for AI output: dangling references, unsupported counts, overclaims.