Skip to main content

Client Integration Guide

This guide covers client integration only: how your application should call KoreShield to secure LLM requests and RAG content. It does not include internal operations or backend implementation details.

Integration Paths​

Choose one of the following:

How It Works​

  1. Your app sends requests to KoreShield instead of the LLM provider.
  2. KoreShield authenticates the request (API key or JWT).
  3. KoreShield applies security policies and logs events.
  4. Safe requests are forwarded to the configured provider.

Core Endpoints (Client-Facing)​

  • POST /v1/chat/completions
  • POST /v1/rag/scan
  • GET /v1/rag/scans and GET /v1/rag/scans/{scan_id}
  • GET /v1/rag/scans/{scan_id}/pack (download request + response bundle)
  • POST /v1/scan
  • POST /v1/scan/batch
  • GET /v1/scans and GET /v1/scans/{scan_id}

RAG scan history is persisted server-side for authenticated users, so you can revisit results and export a full scan pack later.

Authentication​

All protected endpoints accept either:

  • Authorization: Bearer <jwt>
  • X-API-Key: ks_...
  • ks_access_token httpOnly cookie

Next Steps​