RAGLight's privacy protection program consists of 3 layers of measures:
- Full-link localization: Work with Ollama/LMStudio to run the model locally and ensure that the raw data does not leave the intranet environment. Disabling the HuggingFace online API avoids the risk of external transmission.
- vector-store encryption: the Chroma database will be
persist_directoryPoints to the encrypted storage volume byVectorStoreConfigCustomizing collection names to avoid information leakage - access control::
- Set strict read and write permissions on local knowledge base folders
- GitHub source change to private repository + SSH authentication (requires code extension)
- Configuring network isolation when deploying in Docker containers
Special Note: The firewall rules for the Ollama service port (default 11434) need to be checked periodically to block unauthorized access.
This answer comes from the articleRAGLight: Lightweight Retrieval Augmentation Generation Python LibraryThe
May not be reproduced without permission:AI productivity tools " How to address the privacy risks of RAGLight's handling of sensitive corporate data?
English 
简体中文 
日本語 
Deutsch 
Português do Brasil