DataChat

Local database search + OSINT project. I mainly made it so I could search messy datasets without writing SQL every time.

It can import CSV, JSON, SQLite and TXT files, turn simple natural-language searches into SELECT queries, and optionally run a deeper OSINT pass on a matching record. Ollama is only used for the final summary, so normal database search works without it.

Run it

Backend:

cd server
pip install -r requirements.txt
python app.py

Frontend:

cd admin
npm install
npm run dev

The API runs on http://localhost:8000 and the frontend on http://localhost:5173.

Data

By default the server looks for datasets in H:\databases. Change DATABASES_PATH in server/app.py if yours live somewhere else.

There is also a small fake dataset for testing:

cd server
python import_demo.py

A few example searches:

JOHN DOE
How many people in Paris?
Find john.doe@gmail.com
List people in 75001

The OSINT mode uses public web sources and can also call a local Ollama model for the report. The defaults are near the top of server/app.py:

OLLAMA_URL = "http://localhost:11434"
OLLAMA_MODEL = "qwen2.5:7b"

This is for datasets and systems you are allowed to work with. The OSINT side can touch personal data, so use it accordingly.

MIT.