.env.example

# Backblaze B2 (required)
B2_S3_ENDPOINT=https://s3.us-west-004.backblazeb2.com
B2_APPLICATION_KEY_ID=your_key_id
B2_APPLICATION_KEY=your_application_key
B2_BUCKET_NAME=your-bucket-name
# B2_PUBLIC_URL=  # optional public URL for your bucket

# LLM + Embeddings: one API key for everything (default: OpenAI)
OPENAI_API_KEY=your_openai_key
# LLM_PROVIDER=openai              # "openai" (default) or "anthropic"
# LLM_MODEL=gpt-4o                 # default for openai; use claude-sonnet-4-20250514 for anthropic
# EMBEDDING_MODEL=text-embedding-3-small

# Optional: use Anthropic for chat (OPENAI_API_KEY still needed for embeddings)
# ANTHROPIC_API_KEY=your_anthropic_key
# LLM_PROVIDER=anthropic
# LLM_MODEL=claude-sonnet-4-20250514

# LanceDB vector store (optional, defaults to s3://{B2_BUCKET_NAME}/lancedb/)
# LANCEDB_URI=s3://custom-bucket/lancedb/

# Document processing pipeline
CHUNK_SIZE=1000
CHUNK_OVERLAP=200
MAX_CHUNKS_PER_DOC=500