A character saying “I remember” is not the same as showing which fact was saved. An editable record gives the user a way to inspect and repair continuity even when the next generated answer is imperfect.
LoveCore’s homepage makes broad memory claims. A second source retrieval of its original app address returned to the public LoveCore website; no memory-management interface was established. Secrets’ documentation is more concrete, but its current distinction between older memories and paid Lorebooks must be preserved.
Identify the screen before following instructions
Secrets’ current Lorebooks overview says Lorebooks are available on Premium and Ultimate, while free companions use the older memory system. Existing paid chats do not automatically convert.
A chat with Open Lorebook already uses a book. A chat offering Upgrade memory can move from the older system. If the interface still shows HIGH, MEDIUM and LOW importance, those are the legacy controls, not lorebook pins or locks.
This matters because an instruction such as “change the importance level” is not a universal answer for every current Secrets account. Follow the guide for the system the chat actually uses.
Maintain one fictional project
Suppose an adult fictional curator, Ellis, is organizing an exhibition. Its opening moves from Friday to Saturday, while the venue remains the west gallery.
A useful record separates the pieces:
| Information | Appropriate purpose |
|---|---|
| Ellis is a curator | Recurring character identity |
| The exhibition opens Saturday | Current fact |
| The original Friday plan was replaced | Event history |
| Confirm the display lighting | Unfinished task |
Secrets’ lorebook documentation separates Cast, World, Relationship, Facts and Chronicle, alongside Premise and Open threads. That structure supports keeping a corrected date separate from the historical reason it changed.
The user’s practical goal is not to preserve every sentence twice. It is to make the current state clear and keep enough history to explain it.
Choose migration deliberately
The upgrade guide describes choices with materially different consequences:
- Create a new Lorebook uses the existing chat’s available memories and history.
- A book from the same memory group can receive this chat’s history, but later changes to legacy memories are not re-imported automatically.
- Choosing another existing book switches the chat to that book without importing the old memory and history.
- Starting from a Universe creates a new copy of the prepared setting without importing the chat’s old continuity.
The distinction prevents a common mistake: expecting every “use this book” action to merge everything. The guide says visible messages can remain even when the historical memory source has changed. A transcript on screen is therefore not proof that its earlier facts were imported into the selected book.
If Ellis’s existing exhibition must be preserved, the import choices are relevant. If the purpose is a fresh alternate storyline, a separate book may be more appropriate than merging incompatible events.
Review the correction after the move
In the finished record, make Saturday the current opening date. Retain the change from Friday only where it helps explain the history. Remove “confirm the date” from unfinished tasks once it is settled.
Secrets’ legacy memory guide also documents natural-language updates, viewing saved memories and deleting incorrect entries. Those are concrete repair routes for a chat still on the older system, rather than a reason to assume every user has the newer book interface.
The migration guide asks users to inspect the resulting sections; it does not promise a verbatim copy of every message. That is a useful operational limit.
What can a reply demonstrate?
Ask Ellis where and when the installation team should arrive. A reply using Saturday and the west gallery shows use of the corrected facts in that response. It does not prove the system will never regress, or reveal precisely which internal mechanism supplied them.
The supported comparison favors Secrets’ documented manageability for this task, not a claim of superior measured memory. LoveCore’s equivalent controls remain unverified here. Neither product was used in a recall experiment, and no private real-world history is needed to evaluate this fictional project.