How jaycoolslm/sci-chat works
Project type and purpose
• From the README and file tree, this is a “Streamlit App Starter Kit” tailored toward building a “sci‑chat” application: a web UI (Streamlit) that lets users chat with a scientific document‑driven LLM.
• Key features implied: simple front‑end (Streamlit), a chat orchestration layer (chat_with_data.py), data ingestion/scraping modules (app/lib/scrape/*.py), and multiple “site” pages (Analytics, ChatBot, Corpus).
File structure
Top‑level:
• app/ – your application code
– Home.py – the Streamlit entrypoint (defines the overall page layout and navigation)
– lib/ – core business logic
• chat_with_data.py – orchestrates conversations, invokes LLMs, ties in scraped data
• scrape/ – connectors to external data sources (DOAJ, ScienceDirect)
– sites/ – page modules, each representing a Streamlit “tab” or route:
• ChatBot.py, Analytics.py, Corpus.py
• requirements.txt, pyvenv.cfg, bin/ – your virtual‑env and dependencies
• test.py – quick script (likely smoke tests or demo)
Detail level and essential elements
• Every page module and library package as its own component
• Do not omit the test harness or virtual‑env/dependencies layer—they impact deployment and reproducibility