Chats history as part of knowledge base | Voters

Chats history as part of knowledge base

Luke Zilioli

Sometimes I need to refer to an old chat, but its not super easy to find it, or even remember that it exists. It would be really cool if I could chat with the AI in typingmind about the rest of my chats within typingmind. Basically RAG for my typingmind chat library.

April 24, 2025

I also needed this and build a  simple TM extension to do just this. Adds a button, opens a search box with past convos, inject the old convo into the chat area with some reference info! https://github.com/basfenix/tm_convoref/blob/main/convo-reference.js
load as Ext via https://cdn.jsdelivr.net/gh/basfenix/tm_convoref@main/convo-reference.min.js 
Hopefully a good PoC for TM team to see how it could work and add natively one day. 
The rag part on entire library is more tricky, that's basically smart short and long term memory and should be approached carefully but would def be amazing to have in TM!

BF got a screencast?

Hey made a quick screen rec https://drive.google.com/file/d/1GXnqMKfSy4STzgclg_hBYtM40-5aCq5n/view?usp=drivesdk

Ionut Georgian Ciobanu

WillemvG

This would be awesome! I was already working on an automated process to get this done, but native would be much better; and idd, Chatgpt's memory function is actually already doing this.

Cameron Upton

Definitely, even starting with a project folder would be a huge help. With chat gpt their memories feature and recall of past chats has being extremely helpful in the last month especially

nice! and kb be sync'd and also support some api so we can manually insert stuff in there! pls expose an api so i can auto add stuff in there programatiically (increases KB size so u make more money)

Ted H

Yes, this would be really helpful. There are workarounds but if this was native to the TypingMind app it would be super. Even better if all chats within a whole folder/project could be used as a knowledgebase for new chats. This would help with context limitations too - as chats get too long the LLM forgets the beginning of the chat, especially when an attachment is introduced into the chat and the number of tokens used goes way up in one message.