The Intel Fusion Engine Panel
Behind the scenes, the fusion engine continuously collects from your sources, removes duplicates, scores each item, and makes a unified feed available across battlespaces (the conceptual side of this is covered in How Intelligence Flows). The Intel panel is where you interact with that machinery directly — manage what you collect from, and work the graded items it produces.
Opening it
Click ⚡ Intel in the header for the full panel, or use the Intel sidebar tab for a quick preview of the most recent items. The panel has two tabs: Data Sources and Intel Feed.
Data sources
The Data Sources tab lists everything the engine collects from. To add one, click + Add Source and provide:
| Field | What it's for |
|---|---|
| Name | A short label for the source. |
| Endpoint | The address to collect from. |
| Type | The kind of source — a web feed, a structured dataset, a geospatial feed, or manual entry. |
| Category | The primary kind of activity it covers. |
| Poll interval | How often to collect from it automatically. |
| Credibility | Your starting estimate of how reliable it is — the engine refines this on its own as your feedback accumulates. |
Each source shows a colour-coded credibility bar (green = high, amber = moderate, red = low — verify independently). You can trigger an immediate collection with the refresh control, edit a source without losing its history, or remove it entirely.
Run log
The bottom of the tab shows recent collection runs — which source, when, the status (ok / error / running), and how many items were found versus newly stored (duplicates are excluded). It's the quickest way to confirm a source is healthy and actually contributing.
Intel feed
The Intel Feed tab is where you browse and act on graded items. Narrow the feed with:
- Search across titles and descriptions
- Category chips to toggle kinds of activity
- A minimum-validity slider to hide low-confidence items
- A source filter to focus on one source
Each item card shows its title, source, category, time, a colour-coded validity score, a corroboration count (how many independent sources reported it), and severity. Expand a card for the full detail and the analyst controls.
Verifying and flagging
Expanding an item gives you the decisions that train the system:
| Control | Effect |
|---|---|
| ✓ Verify | Marks the item analyst-verified, lifting its validity toward certainty. |
| ✗ Flag false | Marks it a false positive and lowers the source's standing. |
| Notes | A freeform annotation kept with the item. |
Verify and flag-false are mutually exclusive — recording one clears the other. This is where the per-item half of the credibility model gets its human input: automation can only take an item so far, and your verification is the rest (see Credibility & Validity).
The feed runs itself. You don't need to operate the engine day to day — it collects and scores in the background. The panel is for tuning what you collect and judging what comes back.
Where to next
- Credibility & Validity — what the validity score and verification do.
- Feeds & Sources — the broader picture of connecting sources.