Q&A agent API

Objective

Agent that answers questions using OpenAI and a local knowledge base.

Project Structure

The project is organized into three main components:

• frontend/ - Streamlit web UI
• backend/ - FastAPI REST API
• ai/ - AI components (agents, tools, vector store)
• src/ - Shared utilities (database, configuration)

What this project does

Q&A agent using:

• LangChain + langchain-openai
• OpenAI GPT-4.1 (or compatible model via API key)
• Optional local knowledge in the knowledge directory

Setup

• Python: 3.11+
• Poetry installed
• Install deps:

  poetry install

• Set OPENAI_API_KEY (via .env or environment variable)

How to run (Backend - FastAPI)

Start the API server:

poetry run uvicorn backend.app:app --reload --host 0.0.0.0 --port 8000

Then open the interactive Swagger UI:

• API docs: http://localhost:8000/docs

From there you can:

• List messages for a chat:

  • Endpoint: POST /chat/get_messages
  • Example body:

    {
      "chat_id": 1
    }

• Ask a question in a chat (with history + routing + KB):

  • Endpoint: POST /chat/answer
  • Example body:

    {
      "chat_id": 1,
      "question": "What do koalas eat?"
    }

• Delete a chat:

  • Endpoint: DELETE /chat/delete?chat_id=1 (e.g. DELETE /chat/delete?chat_id=1)

How to run (Frontend - Streamlit UI)

Start the Streamlit web interface:

poetry run streamlit run frontend/app.py

The UI will open automatically in your browser at http://localhost:8501.

Features:

• 💬 Chat Interface: Ask questions and get answers from the agent
• 📚 Knowledge Base Management: Upload .txt files to the knowledge base
• 📝 Chat History: View and manage multiple chat sessions
• 🔍 Smart Search: The agent automatically searches the knowledge base when needed