LamChingFung-2425-Anti-FraudX

Feature: Offline Mode Checks

Category: Secondary

Version: 4.1

Last Updated: 2026-03-11

---

Overview

When running with Ollama (local mode), the platform performs startup checks to verify that no data leaves the local network. This is critical for deployments where patient data or sensitive conversation content must remain on-device.

---

Checks Performed

File: backend/utils/offline_mode.py

Function	What It Checks
check_offline_mode()	Verifies GEMINI_ENABLED=false
check_data_isolation()	Confirms no external API keys are configured
verify_ollama_local_only()	Confirms OLLAMA_BASE_URL points to localhost or 127.0.0.1

All three are called from main.py at startup:

check_offline_mode()
check_data_isolation()
verify_ollama_local_only()

Failures log a warning but do not stop the server (non-fatal by design).

---

When Checks Are Skipped

When GEMINI_ENABLED=true, these checks are skipped entirely — data is intentionally sent to the Google Gemini API and the offline guarantee does not apply.

---

Relevant Files

backend/utils/offline_mode.py   All three check functions
backend/main.py                 Calls checks at startup
backend/.env                    GEMINI_ENABLED, OLLAMA_BASE_URL

---

Related Features

• llm-backend.md — Ollama local adapter (the mode this protects)
• gpu-detection.md — Also runs at startup alongside offline checks