Skip to content
SQA Cockpit

Probes

One headline index per system under verification.

live api

Metaintro Chat

live

JSIJobseeker Surface Index

70

Would a real jobseeker have a good time on this surface?

RecognitionSpecificityReachabilityRecencyMatch qualityHostility filterSalary surfacePath coherence
view axes →

Snappy

mock

CQICorpus Quality Index

90

Is the activated corpus complete, fresh, and trustworthy? Mirrors snappy-api's GOLD/GREEN/YELLOW/RED quality bands.

CompletenessFreshnessExtraction qualitySnapshot integrityClassification accuracyCoverage
view runs →

Kai

mock

MRIMemory Recall Index

75

Does kai recall the right memory, ground its answer, and abstain when it should? Built on hybrid dense+BM25+RRF retrieval.

Recall@kGroundednessHybrid relevanceMulti-hop accuracyRetrieval latencyAbstention
view runs →