Skip to content

Settings Reference

Settings are reached from the app’s Settings section and are stored in the database (not in .env), so they survive container restarts as long as data/ is preserved.

Per-service credentials for the integrated OSINT/threat-intel providers (VirusTotal, Shodan, Hunter.io, GitHub, etc.) — see External Services for the full list across every feature. Keys are encrypted at rest before being written to the database; see Security → Secrets at rest. Three providers (VirusTotal, Shodan, Hunter.io) additionally expose a live quota panel here, since those are the only three whose APIs report remaining quota.

Configures which LLM model each AI-assisted feature uses:

  • Default model — the fallback used by any AI feature with no more specific override.
  • Per-feature overrides — newsfeed article analysis, newsfeed report generation, email analysis, AI templates, and photo geolocation can each pin a different model, or leave it unset to fall back to the default.

Available models come from whichever LLM providers are configured (see AI / LLM Providers).

UI-level preferences: dark/light theme, UI language, and command-palette behavior — whether a single-match search result auto-opens, which screen the app opens on (/ route), and whether results always render as tiles instead of tabs.

Enable or disable individual feature modules — a disabled module disappears from the sidebar and command palette without needing a rebuild or restart.

A watch-list of free-text keywords. When enabled, newsfeed articles are scanned (case-insensitive substring match against title, summary, and full text) and any hits are recorded per article as matches, so you can spot articles relevant to your organization or interests at a glance without reading every headline.

A free-form description of your organization’s threat-intelligence profile — profile name, threat sources you care about, indicators of interest, and a minimum severity threshold. This text is rendered to Markdown and injected as context into the AI prompts used for newsfeed article analysis and MITRE ATT&CK enrichment, so the AI’s relevance judgment is tailored to your organization instead of generic.