case.yaml.example

# CaseStack case configuration
# Copy this to case.yaml and customize for your document set.

name: "My Case"
slug: "my-case"
description: "Description of this document collection"

# Where raw documents live (PDFs, images, audio/video)
documents_dir: "./documents"

# Processing options
ocr:
  backend: "pymupdf"        # "docling", "pymupdf", or "both"
  workers: 4

entities:
  spacy_model: "en_core_web_sm"
  types: ["PERSON", "ORG", "GPE", "DATE", "MONEY", "PHONE", "EMAIL_ADDR"]
  # Optional: path to a JSON person registry for fuzzy matching
  # registry: "./data/persons-registry.json"
  # fuzzy_threshold: 85

dedup:
  threshold: 0.90

# Bates number prefixes for dedup overlap detection (optional)
# bates_prefixes: ["EFTA", "DOJ"]

# Serving
serve:
  port: 8001
  title: "My Case — Document Database"
  # ask_proxy:
  #   enabled: true
  #   openrouter_api_key_env: "OPENROUTER_API_KEY"