# Database
DATABASE_URL=postgresql+asyncpg://whatisup:whatisup@localhost:5432/whatisup

# GitHub PAT — fallback / local scripts only. Production collection uses each owner's OAuth token.
GITHUB_TOKEN=

# GitHub OAuth App (callback must be {PUBLIC_APP_URL}/auth/github/callback)
GITHUB_CLIENT_ID=
GITHUB_CLIENT_SECRET=

# Session cookie signing
SESSION_SECRET=change-me-to-a-long-random-string

# Fernet key for GitHub access tokens:
#   python -c "from cryptography.fernet import Fernet; print(Fernet.generate_key().decode())"
TOKEN_ENCRYPTION_KEY=

# Public origin used to build the GitHub OAuth callback.
# Production (Vercel): https://whatisupp.vercel.app
# GitHub OAuth App callback must be {PUBLIC_APP_URL}/auth/github/callback
PUBLIC_APP_URL=http://localhost:8000
FRONTEND_ORIGIN=http://localhost:5173
COOKIE_SECURE=false

# Vercel (Project Settings → Environment Variables): BACKEND_URL = API origin, no trailing slash.
# Do not commit the backend host or IP.

# Optional Chrome extension origin, e.g. chrome-extension://<id>
CHROME_EXTENSION_ORIGIN=

# Cron secret for POST /internal/run-pipeline
CRON_SECRET=change-me-cron-secret

# OpenRouter — get free key at https://openrouter.ai/keys
OPENROUTER_API_KEY=
OPENROUTER_MODEL=meta-llama/llama-3.3-70b-instruct:free

# Admin secret — local POST /admin/run-pipeline
ADMIN_SECRET=change-me-to-something-random

# Collector debounce (seconds)
COLLECT_MIN_INTERVAL=900
# Parallel GitHub users per collect (stay under secondary rate limits)
COLLECT_CONCURRENCY=8
# Public event pages per person (GitHub caps ~300 events)
COLLECT_EVENT_PAGES=3