A repo for example scripts that talk to documentation of any kind of the pattern: Ingest the source (typically public web documentation or PDFs) using Langchain, output a local vector-store with Langchain, then expose a chat window for "QA" using Streamlit UI. "RAG" (Retrieval Augmented Generation) queries against your configured LLM provider, see datascience's architecture diagram in the assets folder.