Can we reconstruct the rules that were in force on a past date?

This is the question that separates an auditable AI deployment from a plausible one. An answer given six months ago was produced under an instruction set that has since been edited, and unless those edits were versioned, nobody can state with confidence what the rules were. Recollection is not evidence.

Skills and prompts live in the organization's own SQL environment as records, carrying version history the same way content does. The Interaction Log ties each execution to the skills and prompt that governed it, so an answer from a past date can be matched to the configuration that produced it. The Prompts and Skills registries show current state as live feeds, which makes the comparison between then and now a query rather than an investigation.