semantra-python

Why we included this project

Semantra turns a folder of local text and PDF files into a searchable collection without you having to stand up any infrastructure. You point the CLI at your documents, it chunks and embeds them on your machine, and a browser interface serves results ranked by meaning rather than keyword overlap. That workflow suits people who regularly dig through large document sets, whether they are journalists combing leaked files, researchers tracing themes across papers, or analysts working through archives. The web app shows relevance scores, highlights the most relevant passages, and lets you refine a query with plus and minus tags. Embeddings run locally by default, with an optional OpenAI-backed mode, so you get search-engine results without sending your documents to a third-party service.