Memory Became a User-Controlled Surface
Why persistent AI memory needs a user-governance layer after the system has already decided what is valid enough to remember.
A memory can be correct and I can still want it gone.
An archive of decisions, failures, changes, and shipped work. The record stays useful by keeping the sequence—and the limits—visible.
Why persistent AI memory needs a user-governance layer after the system has already decided what is valid enough to remember.
A memory can be correct and I can still want it gone.

An analysis of why agentic interfaces need an explicit coherence contract between the state shown to the user and the state an agent can actually address, inspect, or recover.

A reconstructed decision note on why ritual, retained history, Chapters, and non-predictive reflection became more important than AI as Liminal Tarot's headline feature.

A practical test for when an AI result can stay in the transcript and when it needs its own identity, state, revisions, permissions, and actions.