RAG-Powered Coding Helper

A Python tool to query API documentation and technical cheat sheets using Retrieval-Augmented Generation (RAG). Ingests Markdown docs, builds a semantic vector store (FAISS), and answers natural language questions with relevant code snippets or explanations.

Project Structure

rag-coding-helper/
├── main.py                # CLI entry point
├── requirements.txt       # Python dependencies
├── README.md              # Project documentation
├── LICENSE                # MIT License
├── .gitignore             # Git ignore rules
├── data/                  # Markdown docs (API, cheatsheets)
│   ├── pandas.md
│   ├── requests.md
│   ├── ...
├── rag/                   # Core RAG logic
│   ├── helper.py
│   ├── retriever.py
│   ├── llm.py
│   └── __init__.py
├── app/                   # (Optional) App entrypoints
│   └── main.py
├── tests/                 # Unit tests
│   └── test_llm.py

Features

• Ingests Markdown documentation and cheat sheets
• Builds a FAISS vector store for fast semantic search
• Supports HuggingFace, OpenAI, or mock embeddings (offline)
• Natural language question answering with code snippet retrieval
• Extensible: add your own docs to data/

Installation

1. Clone the repository

   git clone https://github.com/yourusername/rag-coding-helper.git
   cd rag-coding-helper

2. Create a virtual environment

   python3 -m venv .venv
   source .venv/bin/activate

3. Install dependencies

   pip install -r requirements.txt

4. (Optional) Set OpenAI API key
   • For OpenAI embeddings, set your API key:

     export OPENAI_API_KEY=your-key-here

Usage Example

Run the CLI and ask a question:

python main.py

Sample session:

Building vectorstore...
Vectorstore ready. You can now ask questions!

Ask a question (or type 'exit' to quit): How do I create a DataFrame in pandas?

Answer:
df = pd.DataFrame({'A': [1, 2], 'B': [3, 4]})
print(df)

Ask a question (or type 'exit' to quit): exit
Goodbye!

Configuration

• Mock Embeddings (default):
  • Works offline, no API keys needed.
  • Set use_mock=True (default in code).
• HuggingFace Embeddings:
  • Install sentence-transformers.
  • Set USE_HUGGINGFACE = True in rag/retriever.py.
• OpenAI Embeddings:
  • Set USE_HUGGINGFACE = False in rag/retriever.py.
  • Set your OPENAI_API_KEY environment variable.

Contributing

• Fork the repo and create a feature branch
• Write clear, well-documented code
• Add or update tests in tests/
• Submit a pull request with a clear description

License

This project is licensed under the MIT License © 2025 ShinyPhoenix000.

References

• LangChain
• FAISS
• HuggingFace Transformers