Local confidentiality chamber: blocks the wrong AI output before it leaves the machine.
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.
Local confidentiality chamber: blocks the wrong AI output before it leaves the machine.