README_PROD.md

FOIA HF Document Search — Production Build

Entry Point

• app.py — orchestrates ingestion, semantic search, export, and audit hooks.

Ingestion Adapters (Present)

• CIA Reading Room
• FBI Vault
• Generic Public FOIA

Missing / Stubbed Adapters (Recommended)

• DoD (incl. components)
• NSA
• DIA
• DHS
• DEA
• ICE

Vector Backend Assumptions

• Current code supports abstract vector ops.
• Recommended backends:
  • FAISS (local)
  • Chroma (persistent)
  • HuggingFace embeddings
  • OpenAI embeddings (optional)

Live Federated Search Upgrade

• Async querying via asyncio + httpx
• Adapter interface with rate limits
• Response caching + deduplication
• Circuit breakers for abuse prevention

Compliance

• Respect robots.txt where applicable
• Rate limiting per agency
• Redaction before export
• Audit logging enabled

Build Timestamp

2026-01-09T23:51:16.728748Z