superS007/localllmjournal: local LLM-powered journaling tool

A personal journaling app powered by a local LLM. Write a brain dump, have an AI-guided conversation to clarify your thoughts, and get a polished journal entry — all running locally on your machine. No cloud, no API keys, your data stays yours.

Brain Dump to Polished Entry

1. Write your raw, unfiltered thoughts
2. An LLM asks gentle clarifying questions to help you reflect
3. When you’re done chatting, it generates a polished journal entry
4. Review, edit, and save

Semantic Search

• Ask natural language questions about your past entries (e.g. “what have I learned about patience?”)
• The app retrieves relevant entries and synthesizes an answer with references

Browse & Manage

• View all entries with mood tags and dates
• Expand any entry to read the full text
• Delete entries you no longer want

• Backend: Python + FastAPI
• Frontend: Vanilla HTML/CSS/JS (no build tools)
• LLM: Ollama (llama3.2:3b for chat, nomic-embed-text for embeddings)
• Storage: SQLite + ChromaDB
• Environment: Conda

# Clone the repo
git clone https://github.com/superS007/localllmjournal.git
cd localllmjournal

# Create the conda environment
conda env create -f environment.yml
conda activate llmjournal

# Pull the required models
ollama pull llama3.2:3b
ollama pull nomic-embed-text

# Run the app
python -m journal.app

Open http://localhost:8000 in your browser.

Brain Dump ──> LLM Dialogue ──> Polished Entry ──> SQLite + ChromaDB
                                                         │
Search Query ──> Embed ──> ChromaDB Lookup ──> LLM Synthesis ──> Answer

The app uses two Ollama models:

• llama3.2:3b — a small, fast chat model that runs well on laptops (including M1 MacBook Air with 8GB RAM)
• nomic-embed-text — generates embeddings for semantic search

All data is stored locally in the data/ directory (SQLite database + ChromaDB vector store).

localllmjournal/
├── environment.yml          # Conda environment
├── journal/
│   ├── app.py               # FastAPI routes + SSE streaming
│   ├── config.py            # Constants and paths
│   ├── db.py                # SQLite CRUD
│   ├── llm.py               # Ollama wrappers
│   ├── brain_dump.py        # Dialogue + polishing logic
│   ├── search.py            # Semantic search pipeline
│   ├── embeddings.py        # ChromaDB integration
│   └── static/              # Frontend (HTML/CSS/JS)
├── prompts/                 # System prompts for the LLM
└── data/                    # Runtime data (gitignored)

MIT