A project page should answer three questions: what was tested, what was observed, and where the claim stops. That is more useful than a triumphant abstract with no method or a repository dump with no interpretation.
This archive therefore separates public conclusions from private working material. Exact implementation versions, plots, raw artifacts, and papers appear when they can be made reproducible and safely shared. Until then, a project still deserves a real description of its scope.
The goal is not to manufacture certainty. It is to leave a map for the work that exists now and a reliable place to attach the material that follows.