UFO·INTEL — embeddable intelligence widget

Live demo

No data source is reachable in the public demo by default — paste your own JSON/RSS endpoint, or click Load sample dataset to use a bundled demonstration fixture so you can exercise every feature offline.

Embed snippet

Replace the script URL below with the deployment URL of your fork (or a relative path if you self-host alongside your portal).

<script type="module" src="https://dascient.github.io/UACP/widget/ufo-intel-widget.js"></script>

<ufo-intel-widget
  data-source="https://your-mirror.example/ufo/index.json"
  refresh-interval="30000"
  theme="dark"
  llm-provider="openai"
  llm-model="gpt-4o-mini"></ufo-intel-widget>

Attributes

• data-source — one or more whitespace-separated URLs (JSON, RSS, or Atom).
• refresh-interval — milliseconds between fetches (minimum 5000).
• theme — dark (default), light, or compact.
• llm-provider — openai, anthropic, or local (OpenAI-compatible).
• llm-model — model identifier passed to the provider.
• api-key — optional inline key (prefer Settings → LLM provider instead).

Privacy

• No telemetry. No external CDN required.
• API key stored only in localStorage on this device.
• The dataset is fetched directly from the source URL you configure and cached in IndexedDB.
• RAG retrieval (TF-IDF) runs in the browser; only the top-K snippets are sent to the LLM.
• An audit log of every refresh, query, and user action is kept locally and exportable.