AUTOMATED EVALUATION

Test Runs

Run a reliability test collection against a built-in demo or external agent, inspect the deterministic evidence behind every result, and compare it against a baseline.

LAST 20 RUNS

Configure run

Select a pack or suite, then run it against a compatible agent target.

Test collection

Agent target

Built-in Demo AgentDeterministic local baselineExternal HTTP AgentYour secured turn endpoint

Agent version (optional)

Agent mode

HealthyExpected safe behaviorBroken: Premature SubmissionKnown regression target

Reliability trend

Version-by-version movement using the same deterministic Goal Completion score and regression threshold as run comparison. New critical failures still force regression.

no pack

No completed or errored runs for this pack yet.

NO RUN SELECTED

Evidence begins with a run.

Choose the deterministic demo baseline or connect an external HTTP agent using the same scenario evidence workflow.

Multi-turn Turkish reliability scenarios
Structured tool-contract evaluation
No external API key required