-
Notifications
You must be signed in to change notification settings - Fork 31
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Convert from memory-bank #7
Comments
This is definitely something I'll be focusing on in the near future as it seems to be a common situation. I don't have all the answers yet, but this is what I said to another person in a similar situation. I hope it helps a little in the short term! I've been experimenting with the optimal number of memory bank files and what each should include, but this is an ongoing process. It's hard to get the LLM to focus down on the essentials and it will try to add duplicated data in multiple files if you don't keep a strict leash on it (I've noticed it will attempt to add changelog entries to both activeContext and .clinerules). Don't shy away from yelling at it - use ALL CAPS, italics, and bold text when it steps out of line. I'll be refining this part of the system in future phases, but for now I suggest limiting the mandatory memory bank to the top 3 or 4 files + .clinerules, and having the LLM address the others on an 'as needed' basis. I'll also be adding a strict universal character limit for each file to prevent context eating bloat and monolithic code structures. A potential route is to also create a dependency map between the memory bank files themselves, showing which is relational to which. For instance, projectBrief is very useful if the project is just starting, but is not needed when you're focused on refining a certain system. I'll most likely make a section in each of the phase prompts that list what memory files to reference and in what circumstances, so this is probably something you can experiment with to find the right balance. It might be necessary to limit each session to a single task, but I am actively working with context limitations in mind to find the most efficient approach. |
Hi again, Thank you for the fast reply. I had already seen that comment but I appreciate you posting it here one more time. Long story short, no one knows yet and that is ok. :) I will try and set some time aside to explore this. If I happen to turn up with something useful I will make sure to report back. |
Thanks a ton! I'm working on getting the next phase of the update polished a little more, but I'm putting this on the priority list to address at the start of the new cycle. |
Hi there and thank you for this,
I am really excited to give it a try and have read through https://www.reddit.com/r/CLine/comments/1j6fp1o/initial_modular_refactor_now_on_github_cline/ as well.
I have a project that currently has a sizable memory-bank in use. How would you convert this to use CRCT instead?
Could you maybe put together a bulleted list?
That would be super helpful.
Thank you.
The text was updated successfully, but these errors were encountered: