I’m announcing a little tool for semantic search of a repository of documents (text, markdown, scripta, minilatex, pdf). This is an experimental project with a lot left
to do, including
- improve the not-so-good UI
- improve the clustering and cluster-naming algorithms
- better documentation!
Installation instructions (github)
The app is made with an Elm frontend and a Python backend. Most of the work was done by Claude AI under pretty strict supervision and instruction. It was a surprisingly smooth and pleasant experience. For example, the last phase of the work was to ask Claude this:
Look up information on how to make a dmg installer (mac), then formulate a plan to make one for this project. Show me your plan before writing any code. (The end result should be a bash script to make the dmg)
Once I told Claude to proceed, the work took about 15 minutes. It required no revision.
I welcome comments and suggestions. This project is both an experiment and a rough cut. I’m sure that there are many things undone, missing, etc.