"Hello there!", in this repositotiry lies a tool for better management of your GitHub stars.
When you teach hundreds of GitHub starred repositories and finding that one specific repository becomes quite the adventure. I have 1.5k...
GitHub's search? waves hand "This is not the search you're looking for."
I needed something more... refined. Something that would bring order to this chaos. Worth mentioning some of these repos are certainly archived or not-maintained as it would be in a perfect world. So here I am, building an elegant and simple solution to tackle this challenge.
There are several options I can do this:
-
A pragmatic way to rely on a knowledge base with natural text search: Notion is a great candidate. Notion AI has some credits in a free plan. Should work just fine in the first iteration.
-
A self-hosted plan: I can have make a system with a simple interface, that can use my natural query language to search for starred repositories and give me a result.
-
A hybrid: I can make a tool customisable, let us say I can locally have a list with details (let say in sqlite db), and then export it to where I want, for instance Notion or custom app.
I will start with option three, and here we outline the basic architecture:
graph TD
subgraph Core ["Core (What do I need now)"]
GH[GitHub API] --> stars-keeper[(stars-keeper)]
stars-keeper --> Notion[Notion Export]
end
By doing so, I think, I have plenty of room for future customisation, supporting different knowledge bases and search engines, that I can use.
- I can fetch a list of all starred repositories
- [x?] I can fetch information about a specific repository
- I understand what is a typical structure of my search in stars looks like
- I can structure this information in a way that I can feed it to LLM and categorize it based on my queries
- I can generate a new draft of star list
- I can use this tool to update current star list
- Golang as a primary language
- ... (more to come, since it's great to set your self boundaries and focus on what matters)