Rule-based recommendations, not black-box AI security claims.
Methodology and trust boundaries
Agent Risk Check maps declared capabilities to practical review items. It does not run your code, scan your live system, or certify your architecture.
Vendor claims are shown with source links, review dates, and confidence labels where available.
No formal audit claim, certification, compliance opinion, or production approval.
No guaranteed best-provider ranking; only providers to evaluate.
Rule library
These rules explain how the tool maps declared agent capabilities to review items. They are methodology notes, not audit certification.
Loading rule catalog…