I identified 35 tools for sharing code with large language models (and created another one ;-). Here’s an overview of open-source tools in this category.
If you’re using a tool that’s not included, please reply to this thread and I’ll add it to the list.
Thank you very much!
I’ve been looking for something like this (not for code, just for markdown text files), but I think I can make it work for me too. I use the Git version control to curate the contents of my Obsidian vaults, which I use for journaling, study and research.
Still, I wish I had more context than 32,000 tokens for 4o. Sadly, this doesn’t seem to be a popular user need (Sam Altman mentioned it briefly in an interview).
Maybe I can find a way to afford a Pro sub in 2025 - that gives you 128,000 tokens of context. Or maybe this context size will eventually trickle down to the Plus tier.
Hi @proxy
I’ve used my tool for writing / research as well as code. It’s pretty much always markdown (or other formats converted first to markdown).
In your case with an Obsidian vault, I imagine that’s lots and lots of small markdown notes? That’s almost perfect for this kind of tool, since inputting them manually is not practical.
Given the context limits, you might be able to filter subsets of notes if they are organized by folder or by name patterns. Our tool allows exclusion and inclusion with .gitignore style globs. You can store sets of these as named profiles (in the configuration files) and switch between them.
Other tools have at least some of these features as well.
Glad to be of help. If you run across a tool you like that I haven’t mentioned, please do post it to this thread.
@proxy, I thought a little bit more about the Obsidian use case. Currently, our program (along with most of the alternatives) adds a folder listing, which shows the entire directory structure of the project (so that the model can ask for missing files).
This makes sense for smaller code projects where this listing is not very large. This has been my own primary use.
The Obsidian case is different because we might have hundreds of very small files. In this situation, it may be better not to add the folder listing, since it’s probably not adding much value and is taking up a lot of context.
WDYT?
I’m happy to explore alternatives as long as someone is willing to validate the usefulness (since I myself don’t have a real use case).