# Ming Salvage Sim environment example
# Copy this file to .env and fill in your own values before running the game.

# Required: OpenAI-compatible Chat Completions API key.
OPENAI_API_KEY=your_api_key_here

# Optional: API endpoint and model.
# Recommended model: deepseek-v4-flash.
OPENAI_BASE_URL=https://api.deepseek.com
OPENAI_MODEL=deepseek-v4-flash

# Optional: per-request timeout for LLM calls, in seconds.
OPENAI_TIMEOUT_SECONDS=180

# Optional: advanced model for the season simulator + score extractor (the two
# heaviest LLM steps each turn). Leave empty to reuse OPENAI_MODEL. Set to a
# stronger / reasoning model (e.g. deepseek-reasoner, gpt-5) for better
# narrative and JSON extraction quality, while ministers/decree/etc keep using
# OPENAI_MODEL to stay cheap and cache-friendly.
# OPENAI_ADVANCED_MODEL=

# Optional: SQLite database path. The app creates this file automatically.
MING_SIM_DB=data/ming_sim.db

# Optional: random seed used for event sampling.
MING_SIM_SEED=7

# Optional: CLI-only overrides. Leave empty to reuse OPENAI_* values.
# CLI_API_KEY=
# CLI_BASE_URL=
# CLI_MODEL=

# Optional: CLI debug start date for a new database only.
# Format: YYYY.MM, for example 1629.04
# MING_SIM_START_YM=

# Optional: automated playtest overrides used by scripts/play_as_emperor.py.
# PLAYTEST_API_KEY=
# PLAYTEST_BASE_URL=https://api.deepseek.com
# PLAYTEST_MODEL=deepseek-v4-flash