ScholarBot

Conversational academic research assistant for the CSUSB library. Students search library resources in natural language via Streamlit; LangGraph manages multi-turn memory; a typed tool searches the CSUSB Primo API.

Stack

• UI: Streamlit
• Workflow / memory: LangGraph + MemorySaver (thread ID)
• Chat LLM: OpenRouter free model (default nex-agi/nex-n2.5-pro:free, with fallbacks)
• Routing decisions: OpenRouter Decisions API + ~typesafe/jev-latest (search vs clarify vs chitchat vs off-topic guardrail)
• Library: CSUSB Primo public Explore REST client

Setup

python -m venv .venv
# Windows
.venv\Scripts\activate
# macOS/Linux
source .venv/bin/activate

pip install -r requirements.txt
cp .env.example .env

Edit .env and set OPENROUTER_API_KEY from OpenRouter keys.

Run

streamlit run app.py

Open the URL Streamlit prints (usually http://localhost:8501).

Docker

Ensure .env has OPENROUTER_API_KEY, then:

docker compose up --build

App: http://localhost:8501

Example conversation

1. "Find recent articles about machine learning" → search + results
2. "Show only books from 2023" → same thread reuses the topic with new filters
3. "Find dissertations by John Smith" → clarifying question for a topic