Interest has risen in tandem with the popularity of LLMs. I have worked on these sorts of things as part of internal tooling.
The idea is that you can provide LLMs with a greater understanding of your codebase to improve code generation results. Additionally, techniques which allow humans to visualize the shape of code are being explored as engineers become less familiar with the specific implementation.
I’m not prepared to speak on the efficacy of the techniques at the moment, but that is the general outline.
pretty crazy stuff. i have been looking at different options to include. does it work on C++ codebases? also do you provide support or plan to maintain it long term? so far i'm looking at some graph based solutions.
btw if you want to see if it can integrate with this architecture https://github.com/ByteAsk/byteask-extensions
Currently I'm working on to implement for C++ codebase and yes I'm going to maintain it for a long term I have also made it open sourced if you want some improvements I can create an issue and then you can contribute to it.
I looked to your product we can work together to integrate both the products!!!
Yeah I am trying to reduce the developer effort in manually finding out the routes in backend or finding out how many files will get impacted if i ran some refactoring of code
16 comments
[ 3.6 ms ] story [ 37.9 ms ] threadonboard init --template clean-architecture
The closest was a "visualization" of a single, unconnected node that can be moved around — all by its lonesome self — on a canvas. For some reason.
why suddenly this becomes popular?
is there a recent breakthough that leads this trend?
The idea is that you can provide LLMs with a greater understanding of your codebase to improve code generation results. Additionally, techniques which allow humans to visualize the shape of code are being explored as engineers become less familiar with the specific implementation.
I’m not prepared to speak on the efficacy of the techniques at the moment, but that is the general outline.
I would think an LLM would be useful for explaining behavior, whereas visualizing code references can be done with no AI at all.
I looked to your product we can work together to integrate both the products!!!