CS Student Copilot 🤖

Developed for the "Intelligent Agents" course at the University of Piraeus, CS Student Copilot is an AI-powered assistant designed to help computer science students with coding, studying, and academic research. It provides a unified interface for code generation and debugging, knowledge base management, and academic paper search—all accessible via CLI or a user-friendly Streamlit web app.

✨Features

The CS Student Copilot is comprised of a coordinator that routes tasks to three specialized agents:

• CodeHelper 👨‍💻

  • Generate, debug, and explain code in Python or other languages.
  • Execute code snippets.
  • Analyze and improve code files or folders.
  • Read and summarize code files.

• StudyBuddy 📖

  • Creates a searchable knowledge base from your local .pdf and .txt files (e.g., lecture notes, e-books, documentation, articles).
  • Ask questions in natural language and get answers synthesized directly from your own documents, complete with source citations.
  • Uses a robust, custom chain to provide reliable and well-formatted answers without unpredictable agent behavior.

• ScholarScout 🎓

  • Uses the Semantic Scholar API to find relevant papers based on keywords, topics, or authors.
  • Retrieves detailed abstracts and "TL;DR" summaries for specific papers.
  • Can download the PDF of any open-access paper directly to a local folder, automatically naming the file with the paper's title.

Quick Start

Prerequisites

• Python 3.11 or higher.
• git for cloning the repository.
• (For StudyBuddy): A local Ollama instance running with an embedding model.

1. Clone the Repository

git clone https://github.com/DimGiagias/cs_student_copilot.git
cd cs_student_copilot

2. Setup Environment and Install Requirements

You can use the provided setup script for easy environment setup (on Linux):

with uv:

./setup.sh --uv

with pip:

./setup.sh

Or use the requirements.txt file directly:

pip install -r requirements.txt

3. Configuration

The application requires an API key from OpenRouter.ai to access free LLMs.

1. Create a .env file in the root of the project directory (an example is provided):

cp .env.example .env

2. Open the .env file and paste your OpenRouter API key:

OPENROUTER_API_KEY="your_openrouter_api_key_here"

Note: For embedding, you need to set up a local Ollama model (see Ollama documentation). We recommend mxbai-embed-large

ollama pull mxbai-embed-large

Usage

🖥️Terminal (CLI)

Run the CLI tool:

python3 main_cli.py

Example Commands

• CodeHelper:

  python3 main_cli.py codehelper "Write a Python function to reverse a string."

• StudyBuddy:

  • Index documents:

    python3 main_cli.py studybuddy index --path ./my_notes

  • Ask a question:

    python3 main_cli.py studybuddy ask "Summarize the main concepts in my calculus notes."

• ScholarScout:

  python3 main_cli.py scholarscout "Find recent papers about transformers in NLP."

🌐Web Interface (Streamlit)

Launch the web app:

streamlit run app_streamlit.py

• Select your assistant from the sidebar.
• Chat with CodeHelper, StudyBuddy, or ScholarScout.
• Manage your knowledge base and view example prompts.

🛠️ Project Structure

• agents/
  Contains the core logic for each agent (code_helper.py, scholar_scout.py, etc.) and the coordinator.py that routes user requests.

• tools/
  Holds the specialized tools that agents use to interact with external services (like the Semantic Scholar API) or the local filesystem.

• rag_components/
  Contains the RAGManager, which encapsulates all the logic for StudyBuddy's Retrieval-Augmented Generation capabilities.

• core/
  Includes project-wide configuration (config.py) and centralized model loading services (llm_service.py).

• main_cli.py
  The entry point for the command-line interface.

• streamlit_app.py
  The entry point for the Streamlit web UI.

• setup.sh
  A simple shell script to automate environment setup.

License

This project is licensed under the MIT License. See the LICENSE file for details.