🌙 SleepBuddy

Live app: https://sleepbuddy-jnkc5xqmpozqeysquaazhw.streamlit.app

Hello, parent! SleepBuddy helps you plan your baby's sleep rhythm based on their age and wake-up time. Get a full day schedule — backed by AAP, WHO and NHS guidelines. Powered by AI.

---

Tech Stack

Streamlit + Claude Haiku API + Google OAuth + Supabase

Component	Technology	Cost
Hosting	Streamlit Cloud	Free
AI model	Claude Haiku (claude-haiku-4-5-20251001)	~$0.001/request
Database	Supabase (PostgreSQL)	Free tier
Auth	Google OAuth 2.0	Free

---

Project Structure

sleepbuddy-app/
├── app.py          # Main Streamlit app
├── auth.py         # Google OAuth
├── db.py           # Supabase — user data read/write
├── ai.py           # Claude Haiku requests
├── knowledge.py    # Knowledge base (AAP, WHO, NHS)
├── limits.py       # Daily usage limit (5/day)
├── requirements.txt
└── .streamlit/
    └── secrets.toml  # API keys (never commit this)

---

Features

• Google OAuth login — user identified by Google email
• Session memory — baby age saved permanently; wake/nap times reset daily
• Daily schedule generation — AI-powered, based on age + wake-up time + naps taken
• Usage limit — 5 AI requests per user per day, auto-reset at midnight
• Admin control — block/unblock users, reset limits via Supabase Dashboard
• Dynamic nap slots — number of nap inputs adjusts based on baby's age (1–5 naps)
• In-app manual — collapsible how-to guide visible after login

---

Setup

1. Supabase

1. Create a project at supabase.com
2. Run this SQL in the SQL Editor:

CREATE TABLE users (
    id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
    email TEXT UNIQUE NOT NULL,
    name TEXT,
    created_at TIMESTAMP DEFAULT NOW(),
    last_seen TIMESTAMP DEFAULT NOW(),
    is_blocked BOOLEAN DEFAULT FALSE,
    baby_age_months INTEGER DEFAULT 8,
    last_wake_time TEXT,
    last_nap1_start TEXT,
    last_nap1_end TEXT,
    last_nap2_start TEXT,
    last_nap2_end TEXT,
    last_nap3_start TEXT,
    last_nap3_end TEXT,
    daily_data_date DATE DEFAULT CURRENT_DATE,
    daily_uses INTEGER DEFAULT 0,
    daily_reset_date DATE DEFAULT CURRENT_DATE
);

3. Copy Project URL and service_role secret key

2. Google OAuth

1. Go to console.cloud.google.com
2. Create project → APIs & Services → Credentials
3. Configure OAuth consent screen (External)
4. Create OAuth 2.0 Client ID (Web Application)
5. Add Authorized redirect URI: http://localhost:8501/callback
6. Copy Client ID and Client Secret

3. Anthropic

1. Go to console.anthropic.com
2. Create an API key

4. Configure secrets

Fill in .streamlit/secrets.toml:

ANTHROPIC_API_KEY = "sk-ant-..."
SUPABASE_URL = "https://your-project-id.supabase.co"
SUPABASE_KEY = "your-service-role-key"
GOOGLE_CLIENT_ID = "xxx.apps.googleusercontent.com"
GOOGLE_CLIENT_SECRET = "GOCSPX-..."
APP_URL = "http://localhost:8501"

5. Run locally

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
streamlit run app.py

---

Admin Controls

Action	Where
View all users	Supabase Dashboard → Table Editor → users
Block a user	Supabase → set is_blocked = true
Reset user limit	Supabase → set daily_uses = 0
Disable Google login	Google Cloud Console → disable OAuth app
Delete a user	Supabase → delete row from users table

---

Cost Estimate (Claude Haiku)

• Input: ~500 tokens = $0.0004
• Output: ~400 tokens = $0.0008
• Total: ~$0.0012 per request

At 5 requests/day per user: $0.006/user/day = **$0.18/user/month**

---

Deployment (Streamlit Cloud)

1. Push code to a private GitHub repo
2. Go to share.streamlit.io → connect repo
3. Settings → Secrets → paste contents of secrets.toml
4. Update APP_URL to your Streamlit Cloud URL
5. Add the Cloud URL to Google OAuth authorized redirect URIs

---

⚠️ Educational tool only. Does not replace consultation with a pediatrician.

---

TODO

• Set up UptimeRobot — free HTTP monitor pinging the app every 5 min to prevent cold starts (1-3 min load time on Streamlit free tier)