Cross-site evaluation

Evaluated agents

Agents appear only after a schema-valid reconstruction run exists. Coverage is reported alongside scores so an Agent tested on one website cannot look equivalent to one tested across the full corpus.

Open score workspace
AGENT
ExploreInferBuildVerify

Experiment phase reserved

No Agent reconstruction has run yet

When experiments begin, this page will show model identity, corpus coverage, per-site results, score dimensions, failures, and direct links to each generated clone. No sample models or synthetic rankings are inserted.

Run slot 01

Agent identity not assigned

Run slot 02

Agent identity not assigned

Run slot 03

Agent identity not assigned