I require Python version 3.10+ for local development.
I use the Google Gemini 2.5 Flash model for AI responses.
I configure a Box Breathing cycle of 4 s Inhale → 4 s Hold → 4 s Exhale → 4 s Hold.
I set the insights auto‑update interval to every 30 days.
I create three Firestore composite indexes (user_id + date, user_id + delivered + deliver_on, user_id + created_date).
To get started locally I create a virtual environment (python -m venv .venv), activate it, install dependencies with pip install -r requirements.txt, populate a .env file with GOOGLE_API_KEY, FIREBASE_WEB_API_KEY, and the full FIREBASE_SERVICE_ACCOUNT_JSON, then run the app via streamlit run app.py. When deploying to Streamlit Cloud I push the repo, then add the same secrets in the platform’s Settings → Secrets page. I never commit the .env or service‑account JSON; instead I rely on Streamlit Secrets for production.