# NWO Design Engine — Environment Variables
# Copy this file to .env and fill in your values

# ── LLM Provider Keys ──────────────────────────────────────────────────────────
ANTHROPIC_API_KEY=
OPENAI_API_KEY=
GEMINI_API_KEY=
MISTRAL_API_KEY=

# Ollama (local models — no key needed, just URL)
OLLAMA_BASE_URL=http://localhost:11434
OLLAMA_DEFAULT_MODEL=llama3

# Generic OpenAI-compatible endpoint (LM Studio, vLLM, Together, Groq, etc.)
OPENAI_COMPATIBLE_BASE_URL=
OPENAI_COMPATIBLE_API_KEY=
OPENAI_COMPATIBLE_DEFAULT_MODEL=

# ── Defaults ───────────────────────────────────────────────────────────────────
# Provider: anthropic | openai | gemini | mistral | ollama | compatible
DEFAULT_PROVIDER=anthropic
DEFAULT_MODEL=                         # Leave blank to use provider default

# Backend: openscad | cadquery
DEFAULT_BACKEND=openscad

# Export format: stl | obj | 3mf
DEFAULT_EXPORT_FORMAT=stl

# ── Paths & Runtime ────────────────────────────────────────────────────────────
OPENSCAD_BIN=openscad                  # Path to OpenSCAD binary
OUTPUT_DIR=./output                    # Where generated files are saved
TEMP_DIR=./tmp                         # Temp workspace for renders

# ── API Server ─────────────────────────────────────────────────────────────────
API_HOST=0.0.0.0
API_PORT=8000
API_WORKERS=4
API_KEY=                               # Optional: protect the API with a key

# ── Validation Thresholds ──────────────────────────────────────────────────────
MIN_WALL_THICKNESS_MM=0.8              # Minimum wall thickness (FDM default)
MAX_OVERHANG_ANGLE_DEG=45.0            # Max overhang without support
AUTO_REPAIR=true                       # Attempt mesh repair on validation fail

# ── Logging ────────────────────────────────────────────────────────────────────
LOG_LEVEL=INFO                         # DEBUG | INFO | WARNING | ERROR
LOG_FORMAT=json                        # json | console