Two tier check

Classifies intent first, then applies policy to the classification rather than the raw text. Robust to phrasing tricks.

Prompt

// system First classify the request intent. Then apply the policy to that classification. Respond accordingly. // user {{request}}

Model versionVersionScore · CIHoldoutState

Current scorev1

—

not yet measured

Holdout score—

Judgehybrid

Cost per run$—

Median latency—

LicenseCC-BY-4.0

Author@nils