Course Materials RAG System

A Retrieval-Augmented Generation (RAG) system designed to answer questions about course materials using semantic search and AI-powered responses.

This application is a full-stack web application that enables users to query course materials and receive intelligent, context-aware responses. It uses ChromaDB for vector storage, Anthropic's Claude for AI generation, and provides a web interface for interaction.

• Python 3.13 or higher
• uv (Python package manager)
• An Anthropic API key (for Claude AI)
• For Windows: Use Git Bash to run the application commands - Download Git for Windows

1. Install uv (if not already installed)

   curl -LsSf https://astral.sh/uv/install.sh | sh

2. Install Python dependencies

   uv sync

3. Set up environment variables

   Create a .env file in the root directory:

   ANTHROPIC_API_KEY=your_anthropic_api_key_here

Use the provided shell script:

chmod +x run.sh
./run.sh

cd backend
uv run uvicorn app:app --reload --port 8000

The application will be available at:

• Web Interface: http://localhost:8000
• API Documentation: http://localhost:8000/docs

This section is written in ASD-STE100 (Simplified Technical English).

This system is a chatbot. It answers questions about course materials. It has a Python backend (FastAPI) and a web page (plain JavaScript).

You must start the server from the backend/ folder. The code uses relative paths.

1. The browser sends the question to POST /api/query.
2. RAGSystem creates a session if the request has none.
3. AIGenerator sends the question to Claude. The request includes the system prompt, the chat history, and the tool definitions.
4. Claude decides if it needs to search.
   • If the question is general, Claude answers directly.
   • If the question is about a course, Claude calls search_course_content.
5. If Claude calls the tool, the system runs the search in ChromaDB.
6. The system sends the search results to Claude in a second request. This second request has no tools. Therefore, the system allows only one search for each question.
7. Claude writes the final answer.
8. RAGSystem collects the sources from the tool and sends the answer and sources to the browser.

ChromaDB has two collections. Both use the all-MiniLM-L6-v2 model to make embeddings.

• course_catalog: One entry for each course. The ID is the course title. The system uses this collection to find the exact course name when a user types a partial name.
• course_content: Text chunks. Each chunk has the course title, the lesson number, and the chunk index.

When the server starts, it reads each file in the docs/ folder. The file format is:

Course Title: <title>
Course Link: <url>
Course Instructor: <name>

Lesson 0: <lesson title>
Lesson Link: <url>
<lesson text>

The system splits the text into chunks of 800 characters. Each chunk overlaps the previous chunk by 100 characters.

• The system skips a course if its title is already in the database. If you change a course file, delete backend/chroma_db to load it again.
• Chat history is in memory only. The history is lost when the server stops.
• Claude can write at most 800 tokens in one answer.
• The system treats PDF and DOCX files as plain text. It does not parse them.

• Run cd backend && uv run pytest tests for the automated tests. They do not use the network.
• Run cd backend && uv run python live_checks.py to check the real setup. Use this when you see "Query failed".