Insights
An insight is something your sources or rules raised: a detection from a connected product, a watcher firing, or a threat-intelligence item. A watcher generates a detection, and the detection lands here as an insight. Insights are browsed on the Insights tab of Activity, where you can see the evidence and entities behind them, follow what they led to, and start a hunt from any one of them. High and Critical ones still waiting for a judgement also appear in the Inbox under Needs you, in Detections to judge.

Filter and search
| Control | Options |
|---|---|
| Time range | Presets such as Last 24 hours, Last 7 days, Last 30 days, or a custom range. |
| Type chips | One chip per insight type present in the window (for example Detection Finding, Threat Intelligence, Vulnerability Finding), with counts. All clears the chip. |
| Origin | Any origin, Watcher, Connection or Manual — who raised the insight. |
| Needs review | Show only insights still waiting for someone to judge them. |
| Search | The header search box (Search activity…) matches titles. |
| Sort | Newest first, Oldest first, Most severe first, Least severe first, Name A–Z, Name Z–A. |
| Per page | 10, 20, 30, 50 or 100 rows. |
Watcher detections land as Needs review and stay there until someone acts. The quickest way through them is Needs you in the Inbox, which shows Scout's draft verdict and lets you decide each one in place. See Judge detections in Needs you.
Severity
Every insight carries a severity, shown as a coloured badge in the list and the panel. Sorting by severity ranks Critical first; insights with no severity sort last.
| Severity | Colour |
|---|---|
| Critical | Red |
| High | Orange |
| Medium | Yellow |
| Low | Blue |
| Informational | Grey |
Severity comes from the source and is not edited in Huntbase.
Review states
Insights carry a review state, shown in a bar under the panel header.
| State | Meaning | Buttons |
|---|---|---|
| Needs review | Nobody has judged it yet. | Record verdict |
| Confirmed | Real and worth acting on. | Hunt this, Change verdict |
| Hunt started | A hunt has been opened from it. Final. | — |
| Dismissed | Not worth pursuing. Final. | — |
When a decision carries a verdict, the bar shows it with the state, for example Dismissed · False positive — Patch window or Dismissed · Benign, with the reviewer's reason. A confirmed insight reads just Confirmed. Record verdict (or Change verdict once it's settled) opens the same verdict picker as Needs you and Conclude: choose Confirmed, Benign or False positive, give a reason (required), and click Record with the verdict's name. The same control is in the Full detail drawer in Needs you, and the insight leaves Needs you as soon as you record it. A settled insight with no open hunt offers no next step; with an open hunt, Open hunt stays the main action. Only the allowed transitions are offered. Reopening an insight clears its verdict and reason.

Judge detections in Needs you
Needs you in the Inbox lists unreviewed High and Critical insights under Detections to judge, beside the hunts and approvals waiting on you. Open one and a card shows the severity and title, Raised by watcher … (linked to the watcher), Scout's draft verdict when it has one, up to three pieces of evidence and the entities involved. Decide with:
| Button | Key | What it records |
|---|---|---|
| Confirm | C | The insight is real: it moves to Confirmed. |
| Benign | B | Real activity, but expected: Dismissed · Benign. |
| False positive | F | The rule fired on the wrong thing: Dismissed · False positive. |
| Hunt this | H | Starts a hunt from the insight. |
The button that matches Scout's draft is the filled one. U undoes the last decision, which puts the insight back to Needs review. You can also select several and apply Benign or False positive to all of them at once; Confirm is always one at a time. Detections from a connected tool and graph-rule detections are judged the same way, with a reason, and Undo reopens them. See Needs you.
The insight panel
Click a row to open the insight in the side panel.
Header. The title (usually the source's rule name), the kind beneath it (for example Detection), and a row of badges: severity, High / Medium / Low confidence where the source supplied one, the raw status, the source product when no watcher raised it, the connection it came through, and how long ago it happened. View in [vendor] deep-links back to the source console when the finding carries a link. The summary icon (Summarise with Scout / Regenerate summary) writes Scout's plain-language read.
Linked. A strip under the header shows where the insight came from and what it led to:
- ⟵ Watcher: name — the watcher that raised it. Click to open the watcher.
- ⟶ the hunts started from it, for example 1 hunt · Confirmed or 3 hunts (1 open). A chip for one hunt opens it; a chip for several opens a list with each hunt's verdict.
When there is neither, the strip says No hunts or watcher linked yet.
Next steps. Under the summary, Start hunt starts a hunt seeded with the insight; when one is already open for it, the button reads Open hunt, with New hunt beside it to start a fresh one. Investigate in graph opens the insight's graph in Explorer, and Ask Scout asks Scout about it — in its hunt when one is open, otherwise in a new chat.
Information tab.
| Section | Contents |
|---|---|
| Summary and description | One line on what happened, to what, and how sure we are, then the source's own description (or a plain-language explanation of the insight type when the source sent none). |
| Subjects | The hosts, users and other entities involved, as entity chips. |
| Vulnerabilities | For CVE-related insights: CVSS, EPSS, CISA KEV / public exploit, and links out. |
| Why this fired | The conditions that matched, where the source or watcher recorded them. |
| Details | Time, source, watcher or connection, severity, confidence, status and class. A detection raised by a watcher also shows Seen in: every place the event was found, with a Store verdict badge where the store decided the match. See One detection, many stores. |
| Evidence | The observables in the finding, grouped by type. Each value is copyable. |
| Tags | Add or remove tags. |
| Identifiers / Additional metadata | Collapsed. The raw IDs and the record as the source sent it, for when you need the exact field. |
Relationships tab. A one-sentence summary of what the insight is directly connected to, then a read-only Neighborhood graph of those entities. Investigate in graph opens the same graph in an Explorer tab. If the source sent an alert without host or file detail, the tab says so instead of showing an empty canvas. See Entities.
Start a hunt from an insight
Click Start hunt in the panel. Huntbase starts a hunt from the insight straight away and opens it; if that can't be done for this insight, Hunt with Scout opens instead with the finding as the seed, so you can add What are you looking for?, adjust the Starting hypothesis and click Start with Scout. An insight named for a CVE drafts a question such as "Was CVE-2021-44228 exploited on Apache Log4j (web-prod-01)?" rather than a generic one. See Start a hunt.
You can also tick several insights (and entities) in the list and click Create hunt in the floating bar to start one hunt seeded with all of them.
Insight types
Insight types (the Type chips and the Class in the panel) are defined by your Huntbase administrators under Admin › Content Management › Insight Types, mapped to OCSF classes.
Next steps
- Watchers — the rules that raise detections automatically
- Inbox — judge detections under Needs you
- Start a hunt — what happens after Start hunt
- Entities — the objects an insight's evidence points at