Feature Idea: A Dedicated “User Rules” Section
Hello OpenAI team,
I’d like to share an idea that actually came to me through my own experience using ChatGPT. I don’t know whether something like this is already being considered or whether it would work exactly the way I imagine it, but I wanted to share the idea because I think it could become a really useful feature.
I use ChatGPT for many things, including very detailed role-playing. Over time, I developed a fairly extensive set of personal rules for how I want my role-plays to work. My current rule foundation, for example, contains 31 individual rules.
These rules are not simply about writing style. They define things such as how the roles between ChatGPT and me are divided, how my original character should be handled, how canon characters should behave, what information individual characters are allowed to know, how the story should progress, how the narrative focus should be handled, which writing conventions should be used, and even what should be checked before an answer is sent.
This made me wonder:
Why do permanent working rules like these have to be handled through Memory or through the context of an individual conversation?
That led me to an idea for a separate feature:
“User Rules”
I imagine a dedicated section where users could create their own individual rules for how ChatGPT should work with them.
I would see this as something different from Memory.
For example:
Memory:
What should ChatGPT remember about me?
User Rules:
How should ChatGPT work with me?
Character Profiles:
Which character information that I created should be treated as binding?
Project/Story Context:
What has happened so far in this particular project?
Current Chat:
What is happening right now?
I think separating these things could make ChatGPT much more reliable and easier to use.
For example, a user could create a rule such as:
“In my role-plays, I have complete control over my own character. Never independently write my character’s thoughts, feelings, decisions, or actions.”
Or:
“When we write stories together, use a detailed and atmospheric novel-like writing style.”
Or:
“When writing fanfiction, canon characters should remain faithful to their original personalities and should only know information they could realistically know within the story.”
These would not really be memories about the user. They would be persistent working instructions for how ChatGPT should collaborate with that user.
I could imagine users being able to create multiple rule sets, for example:
Role-playing
Creative writing
Fanfiction
Programming
Learning
Applications- or completely custom rule sets
Users could then activate or deactivate these rule sets or assign them to specific projects.
Character Profiles
Another feature that I think would fit extremely well into this system would be persistent user-created character profiles.
For role-playing and creative writing, character profiles can become very extensive. They can contain a character’s name, appearance, personality, abilities, background, relationships, and many other details.
A user could create a character profile and mark it as:
“This character profile is binding and must not be independently changed, expanded, reduced, or contradicted by ChatGPT.”
Then, when starting a new role-play, the user could simply select the appropriate character profile.
This would be much cleaner than trying to manage all of this information through ordinary memories.
Especially useful for long role-plays
For me personally, this would be incredibly useful when a role-play becomes so long that I have to continue it in a new chat.
At the moment, moving a very long role-play to a new conversation can mean having to re-establish many of the rules and instructions again.
With a permanent User Rules system, the rules would already be available.
The character profile would already be available as well.
I would then only need to provide a summary of what has happened so far:
“This is what has happened up to this point…
This is the current situation…
This is where we are continuing from…”
The permanent rules and the actual story history would therefore be separated.
I think this could also make much better use of the conversation context, because the context would not have to carry the entire rulebook just to remind ChatGPT how the role-play is supposed to work.
Writing conventions could benefit from this too
For example, my own role-playing rules include specific writing conventions.
Actions are written using *…*, thoughts using •…•, memories and flashbacks using //…//, remembered words using >…<, and so on.
These conventions are intended to be used by both me and ChatGPT.
With a normal conversation context, however, it is possible for such rules to become less prominent over time or not always be applied consistently.
If they were stored as active User Rules instead, they would not simply be information from an earlier conversation. They would be an explicit working instruction for the active rule set.
Taking the idea even further
I could even imagine User Rules being structured rather than simply being one large text box.
For example:
Rule 1 – OC Control
Scope: Role-playing
Status: Binding
Rule 2 – Narrative Focus
Scope: Role-playing
Status: Binding
Rule 3 – Writing Style
Scope: Role-playing and creative writing
Status: Binding
Users could edit, reorder, enable, or disable rules and potentially choose whether a rule applies globally, to a specific project, or only to one conversation.
For complex rule sets, perhaps ChatGPT could also internally check whether a planned response is consistent with the active User Rules before sending it.
I realize that the technical implementation is probably much more complicated than it sounds from a user’s perspective. 
I’m not a developer, so I’m certainly not trying to tell you how such a system should technically be built.
I simply wanted to share the idea because I encountered this problem through actually using ChatGPT, and that experience led me to this concept.
I also don’t think this would only be useful for role-playing.
People use ChatGPT in very different ways. Some people write stories, work on long-term projects, program, study, write fanfiction, develop fictional worlds, or use ChatGPT for highly specialized workflows.
Each of these use cases can have very different requirements for how ChatGPT should work with the user.
That is why I think a persistent, individually configurable system for personal working rules could be really interesting.
Maybe this is an idea worth considering for a future version of ChatGPT.
Thank you for taking the time to read user feedback and ideas. 