# Groq chat model (required). Never put this key in frontend code. # Get a key at https://console.groq.com/keys GROQ_API_KEY= GROQ_MODEL=openai/gpt-oss-20b # MiniLM embeddings use ZeroGPU on HF and CPU locally. # Changing embeddings requires clearing and re-indexing existing documents. # DENSE_MODEL=Qwen/Qwen3-Embedding-0.6B # Optional: point at a persistent disk mount (e.g. /data in Docker). # DATA_DIR=/data # Optional: protect the shared document library and Groq quota with a login. # Set both values, preferably as hosting secrets. # GRADIO_AUTH_USERNAME= # GRADIO_AUTH_PASSWORD= # Langfuse Observability (optional) # Set LANGFUSE_ENABLED=true and provide your API keys to enable tracing. # Get your keys from https://cloud.langfuse.com or your self-hosted instance. LANGFUSE_ENABLED=false LANGFUSE_PUBLIC_KEY=pk-lf-... LANGFUSE_SECRET_KEY=sk-lf-... LANGFUSE_BASE_URL=http://localhost:3000