Settings

AI Provider

Choose which AI engine powers your conversations. Open Settings > AI Provider to switch between providers:

• llama.cpp — Built-in engine, download and run GGUF models locally
• Apple Foundation Models — Apple's on-device AI (requires macOS 26+)
• Ollama — Connect to an existing Ollama installation on your Mac

After selecting a provider, choose a default model from the available models for that provider. New chat sessions will use this model automatically.

Models

Manage your AI models from Settings > Models:

• Download new models — Browse available GGUF models and download them directly within the app
• View existing models — See all downloaded models with their size and format
• Delete models — Remove models you no longer need to free disk space

Models are stored in the app's sandboxed container. See AI Models for detailed guidance on choosing the right model.

Encryption

SafeRag SE supports AES-GCM encryption for your chat messages at rest. When enabled, messages are encrypted before being stored in the local database.

• Enable encryption — Turn on encryption in Settings and set a passphrase
• Change passphrase — Update your encryption passphrase at any time
• Disable encryption — Turn off encryption for future messages (existing encrypted messages remain encrypted)

Documents

Configure how SafeRag SE handles your documents for RAG:

• Watch folders — Set up folders that SafeRag SE automatically monitors for new documents
• Folder access — Grant folder permissions when prompted. Because SafeRag SE runs in a sandbox, it needs your explicit permission to access folders outside its container.
• Processing settings — Configure how documents are chunked and indexed for vector search

See RAG & Documents for more on working with documents.

Prompt Templates

Save frequently used prompts as templates so you can reuse them quickly:

• Create templates — Write a prompt and save it with a descriptive name
• Edit templates — Modify existing templates to refine your prompts
• Delete templates — Remove templates you no longer need
• Template variables — Use placeholders in your templates that get filled in when you use them

Appearance

SafeRag SE follows your macOS system appearance settings. The app automatically adapts to Light and Dark mode based on your System Settings.

Data Management

Manage your local data from the Settings panel:

• Clear chat history — Delete all chat sessions and messages
• Export data — Export your conversations for backup or transfer

Data Storage Location

All SafeRag SE data is stored locally in the app's sandboxed container:

~/Library/Containers/io.corixa.SafeRagSE/

This includes the database, downloaded models, indexed documents, and settings. No data is stored on external servers.

About

The About section in Settings shows:

• Version — Current app version and build number
• Acknowledgments — Open-source libraries used by SafeRag SE
• Links — Quick access to the product page, documentation, privacy policy, and support