System-state preservation
What state must survive so later review does not depend on memory and fragmented logs?
Research
Research explores lineage without surveillance, third-party verification without platform dependence, evidence sufficiency, bounded automation, and reproducible technical state. Research remains separate from released product claims.
What state must survive so later review does not depend on memory and fragmented logs?
How can reviewers test supported artifact claims without trusting a single interface?
How should scope, approvals, context, evidence, and stop conditions constrain agent action?
How can people prepare for evidence preservation before an incident or dispute?
How can owned devices and controlled laboratories preserve reproducible technical state?
What separates a pile of artifacts from a bounded, reviewable evidence package?