REALM: Personal Notes + LLM Query System

REALM is a personal knowledge management system that connects your Personal notes to a local vector store and lets you query your notes using a Large Language Model (LLM). It supports automatic syncing, embedding with Chroma DB, and querying with Gemini via LangChain.

---

🚀 Features

• Syncs only updated or new Markdown files from your Personal notes
• Stores embeddings locally using ChromaDB
• Queries your notes using Gemini (Google Generative AI)
• Command-line interface for seamless access

---

📁 Project Structure

myBrain/
│
├── embeddings/              # Stores Chroma vector database and metadata
├── scripts/
│   ├── create_embeddings.py # Sync and embed vault notes
│   └── langchain_integration.py # LLM querying logic
├── .env                     # Stores environment variables (DO NOT COMMIT)
├── brain.sh                 # CLI entrypoint script
├── requirements.txt         # Python dependencies
└── README.md

---

⚙️ Setup

1. Clone the Repo

git clone https://github.com/yourusername/myBrain.git
cd myBrain

2. Create a Python Virtual Environment

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

3. Install Requirements

pip install -r requirements.txt

4. Create a .env File

VAULT_PATH=/absolute/path/to/your/obsidian/vault
GEMINI_API_KEY=your_google_gemini_api_key

Run the script

Now you can run the script anytime

./brain.sh

This will:

1. Activate the virtual environment
2. Sync and embed updated notes from your vault
3. Start the LLM-powered QA system
4. Prompt you to ask questions about your knowledge base
5. Type exit to quit