Select intent to auto-configure evaluation modes.
Used for all runs below.
Profile
Legal — Refusal-Strict. High compliance + strict refusals. Use for legal/policy/regulated.
Schema/JSON/Drift endpoints currently use server defaults.
New here? Open Demo Library →
Compare → Baseline vs Enforced output comparison
Drift → Repeat runs to measure variance
Drift Delta → Variance reduction (baseline vs enforced)
Schema → Structured output validation
JSON Strict → Strict JSON compliance check
Demo prompt library
Click “Run” to load the prompt, select recommended mode(s), switch to Run, and execute.
Results
Latest run outputs. Use “View” for the modal viewer.
Recent Runs
Signed-in only. Shows latest usage rows and allows bundle download.
Account
Usage + entitlements (requires sign-in). No JWT is shown.
Proof Pack
Read-only artifacts (cost / stability / refusal precision).