.env.example

# debuga.ai LLM Stack - Environment Variables (Example)
# Copy this file to .env and fill in your values.
# NEVER commit .env with real values.

# ─── HuggingFace ───────────────────────────────────────────
# Required to download gated models from HuggingFace Hub
HF_TOKEN=your-huggingface-token-here

# ─── vLLM Engine ───────────────────────────────────────────
# Model to load (HuggingFace model ID)
VLLM_MODEL_ID=Qwen/Qwen2.5-Coder-7B-Instruct

# GPU configuration
VLLM_TENSOR_PARALLEL_SIZE=1
VLLM_GPU_MEMORY_UTILIZATION=0.90
VLLM_MAX_MODEL_LEN=8192

# Server
VLLM_HOST=0.0.0.0
VLLM_PORT=8000

# ─── LLM Gateway ──────────────────────────────────────────
# Feature flag: enable local inference routing
ENABLE_LOCAL_INFERENCE=false

# Provider URLs
CLOUD_PROVIDER_URL=http://your-cloud-llm-api/v1
CLOUD_PROVIDER_KEY=your-cloud-api-key-here
VLLM_URL=http://localhost:8000/v1

# Routing strategy: cloud | local | auto
PREFERRED_PROVIDER=cloud

# Gateway server
GATEWAY_PORT=3100

# ─── Monitoring (optional) ─────────────────────────────────
PROMETHEUS_PORT=9090
GRAFANA_PORT=3200