Settings
The Settings page controls instance-wide preferences and integrations. It is organised into three tabs:
| Tab | What it controls | |---|---|---| | General | Instance locale (language, timezone, time format), AI assistant toggle and default model, Hugging Face token | | AI Providers | API credentials for LLM providers used by the AI assistant and AI-powered custom detectors | | MCP Server | MCP endpoint configuration, bearer tokens, and client setup guides |
General settings
The General tab manages how the UI displays content and whether AI assistant features are enabled.
Language, timezone and time format
These three settings each support two modes: automatic (detected from the user’s browser) or a fixed value set at the instance level.
| Setting | Options | Auto-detection |
|---|---|---|
| Language | Automatic, English, German | Browser language |
| Timezone | Automatic or any IANA timezone | Browser timezone |
| Time format | Automatic, 12-hour, 24-hour | Browser hour cycle preference |
When Automatic is selected, the system uses the browser’s built-in capabilities to detect the user’s preference — no IP geolocation or server-side detection is involved. A user can also override the language for their session via a cookie without affecting the instance default.
See Instance Settings for full details.
AI assistant
The AI assistant toggle enables or disables AI-powered features across the entire instance. When enabled, you must select a default AI provider and model that the assistant uses for all AI interactions. This default provider is configured separately on the AI Providers tab.
Hugging Face token
A Hugging Face token authenticates model downloads from the Hugging Face Hub. It is required for audio transcription and Hugging Face custom detectors — without it, downloads fall back to unauthenticated requests with lower rate limits and slower speeds.
The token can be provided in two ways: instance-level (via the Helm chart on Kubernetes deployments) or user-configured (via the Settings UI). When an instance-level token is set, the UI field is disabled and user tokens are ignored.
See Instance Settings for full details.
AI Providers
AI Providers store the API credentials and model configuration for the LLM providers that power the AI assistant and custom AI detectors. Classifyre supports three provider types:
- OpenAI-Compatible — any API that follows the OpenAI chat completions format (OpenAI, Azure OpenAI, Together AI, OpenRouter, Groq, etc.)
- Claude (Anthropic)
- Gemini (Google)
Each provider configuration includes the model identifier, API key, optional base URL (for OpenAI-compatible endpoints), context size, and vision support flag. A built-in test connection button validates credentials before saving.
See AI Providers for configuration reference.
MCP Server
The Model Context Protocol (MCP) server exposes Classifyre’s data and operations to AI clients like Claude, Cursor, and Codex CLI over a standardised HTTP API. Each client authenticates with a unique bearer token.
See MCP Server for connection setup, client configuration guides, capability reference, and troubleshooting.