.env.example

ENVIRONMENT=development
POSTGRES_URL=postgresql+asyncpg://worldcup:CHANGE_ME@localhost:5432/worldcup
REDIS_URL=redis://localhost:6379/0

# ============================================================
# Data source API keys — sign up at each provider's website
# ============================================================

# football-data.org: https://www.football-data.org/
FOOTBALL_DATA_API_KEY=
# Event Registry: https://eventregistry.org/
EVENT_REGISTRY_API_KEY=
# GDELT: https://www.gdeltproject.org/
GDELT_BASE_URL=https://api.gdeltproject.org/api/v2/doc/doc

# The Odds API: https://the-odds-api.com/
ODDS_API_KEY=

# Market odds providers — two different services, use the one you have
# API-Sports / api-football.com: https://v3.football.api-sports.io  (auth=x-apisports-key header)
API_FOOTBALL_KEY=
# apifootball.com: https://apiv3.apifootball.com/  (auth=APIkey query param)
APIFOOTBALL_COM_KEY=

# ============================================================
# LLM configuration — DeepSeek (OpenAI-compatible API)
# IMPORTANT: Do NOT include /v1 in the URL.
# The OpenAI SDK auto-appends /v1, so /v1 here causes double /v1/v1.
# ============================================================
LLM_PROVIDER=deepseek
LLM_BASE_URL=https://api.deepseek.com
LLM_API_KEY=
LLM_MODEL=deepseek-v4-pro
LLM_TIMEOUT_SECONDS=30

ADMIN_TOKEN=change-me
CORS_ORIGINS=http://localhost:5173,http://127.0.0.1:5173
SENTRY_DSN=
VITE_SENTRY_DSN=
MODEL_ARTIFACT_DIR=backend/model_artifacts
PREDICTION_MODEL_VERSION=4.7.0-alpha
EMBEDDING_MODE=local

VITE_API_BASE_URL=http://localhost:8000
VITE_USE_MOCK_DATA=false