Journal Chat

A fully local, private RAG CLI for chatting with your personal journal entries.

Prerequisites

• Python 3.11+
• Ollama installed and running

Ollama Models

Pull the required models before first use:

ollama pull mxbai-embed-large
ollama pull qwen3.5

Setup

python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Usage

1. Prepare your journal

Place your Day One JSON export (or compatible format) in the project directory. The default expected file is journal.json.

The expected JSON structure:

{
  "metadata": { "version": "1.0" },
  "entries": [
    {
      "creationDate": "2024-01-15T10:30:00Z",
      "text": "Journal entry text...",
      "tags": ["tag1", "tag2"],
      "location": {
        "placeName": "Place",
        "localityName": "City",
        "administrativeArea": "State",
        "country": "Country"
      },
      "weather": {
        "conditionsDescription": "Sunny",
        "temperatureCelsius": 25
      }
    }
  ]
}

2. Ingest your journal

python main.py --ingest

To use a custom file path:

python main.py --ingest --file /path/to/your/journal.json

This reads the JSON, extracts text and metadata (date, location, tags, weather), chunks the content, and stores embeddings in a local ChromaDB vector database.

3. Chat with your journal

python main.py --chat

Type your questions and the LLM will answer based on relevant journal entries. Type exit or quit to stop.

Configuration

The following constants in main.py can be adjusted:

Constant	Default	Description
EMBEDDING_MODEL	mxbai-embed-large	Ollama embedding model
CHAT_MODEL	qwen3.5	Ollama chat model
DB_DIR	./chroma_db	ChromaDB storage directory

Privacy

Everything runs locally:

• Ollama runs models on your machine
• ChromaDB stores vectors on disk
• No network calls are made during ingestion or chat