# Using it

> Use the Whisper Splunk add-on: verdicts on your events with whisperlookup, infrastructure context with whisperenrich, and the Investigate dashboard.

*Source: https://www.whisper.security/docs/integrations/splunk/using-it*
*Published: 2026-08-09*
*Last updated: 2026-11-08*

---
Three jobs cover most of a shift: a verdict on every indicator in your results, who runs the ones you are investigating, and one indicator opened on the Investigate dashboard. Every `whisperlookup` row carries what Whisper knows and whether the call completed; [Coverage and call status](/docs/integrations/splunk/coverage-and-status) explains how to read the two.

> **Read `coverage` before `band`.** Only `known-clean` licenses the word "clean"; `no-data` means
> *unknown*, which is a different thing again; `malicious-evidenced` and `ambiguous` mean there is
> evidence, whatever the band says.
> Full contract: [Coverage — what we looked at](/docs/whisper-graph/procedures/coverage).

## Search commands

### Put a verdict on your events

`whisperlookup` reads the field you name, works out whether each value is a hostname, an IP address or a URL, and adds a verdict to each event. Aggregate first, so each indicator is asked about once:

```spl
index=proxy earliest=-24h
| stats count by dest_host
| whisperlookup field=dest_host
| `whisper_marks`
| sort whisper_coverage_rank
| table dest_host count whisper_coverage_mark whisper_status_mark whisper_band whisper_asn_name
```

![whisperlookup over five indicators, shown through whisper_marks: each row says what Whisper knows, from Evidence of malice to No data, whether the call was answered, the band and the network behind the indicator.](/media/splunk-2.0.0-whisperlookup-triage.png)

The `whisper_marks` macro turns the two status fields into words, such as "Evidence of malice" and "Answered", and adds `whisper_coverage_rank` so you can sort the worst first.

- `mode=context`, the default, adds who runs each indicator and its network. `mode=verdict` returns the verdict alone.
- `evidence=true` adds what each verdict rests on: `whisper_evidence`, `whisper_sources` and `whisper_factors`.
- `type=domain`, `type=ip` or `type=url` overrides the automatic choice.
- A search checks up to 5,000 distinct values; `max_distinct` changes that, and values past it come back `not-checked`.
- Answers are cached for 24 hours. `use_cache=false` asks Whisper again.

### Add who runs an indicator

`whisperenrich` answers what an indicator is and what it connects to: its network and prefix, the organisation behind it, its nameservers, how many names share its address and the feeds that list it. It costs more per indicator than `whisperlookup`, so run it on the events you are investigating:

```spl
index=proxy earliest=-4h src_ip="10.1.2.3"
| stats count by dest_host
| whisperenrich field=dest_host
| table dest_host count whisper_asn_name whisper_prefix whisper_nameservers whisper_cohost_count whisper_feed_names
```

![whisperenrich on one hostname: its threat level, the name of the network that routes it, the prefix, how many names share its address, and the feeds that list it.](/media/splunk-2.0.0-whisperenrich-context.png)

`detail=full` adds every field the graph walk produces. For a hostname, `whisperenrich` reads `whisper_is_threat` from the addresses the name resolves to; for the name's own verdict, use `whisperlookup`.

### Ask the graph directly

Seven macros answer the usual next questions about one indicator without Cypher:

```spl
| `whisper_takeover_check("shop.example.com")`
```

[Reference](/docs/integrations/splunk/reference#macros) lists each macro and what it returns. For anything else, `whisperquery` runs your own read-only Cypher, and `whisperschema` describes what the graph holds:

```spl
| whisperquery query="MATCH (h:HOSTNAME {name: $domain})-[:RESOLVES_TO]->(ip:IPV4) RETURN ip.name LIMIT 10" params="domain=example.com"
```

A query with a write clause comes back as one `not-checked` row and is never sent. [Cypher](/docs/cypher) documents the query language.

## Investigate one indicator

Open **Investigate** from the add-on's menu and enter a hostname, IP address or URL. From any search, the field menu on `src_ip`, `dest_ip`, `query` and `url` opens a value there too, in a new tab.

![The Investigate dashboard's Verdict panel for a malicious hostname: Evidence of malice, band HIGH, and the call answered.](/media/splunk-2.0.0-dashboard-verdict.png)

The panels read top to bottom:

- **Verdict**: what Whisper knows about the indicator, its band, and whether the call was answered.
- **The call, and the verdict's parts**: the band, Whisper's own label, the feed count and the answer's age, plus the reason and the fix when the call did not complete.
- **Why — the verdict's own basis**: the lists the indicator appears on, popularity lists and allowlists included, and the factors behind Whisper's score.
- **The network**: the prefix and network that route it, the organisation that runs that network, and the country the network is registered in.
- **The neighbourhood, assessed in one call**: the other names on the same address, counted by what Whisper knows. Above 100 neighbours the panel summarises the shared address instead. With your web or DNS index set on the Settings tab, a column counts how many appear in it.

![The verdict's basis, with the lists and factors behind it, above the network that routes the hostname and the country that network is registered in.](/media/splunk-2.0.0-dashboard-evidence.png)

An indicator Whisper holds no record of shows **Assessment: no infrastructure record**, with "No data" as what Whisper knows. The call was answered and the indicator is unknown, which is not the same as clean.

![The Investigate dashboard for an indicator Whisper holds no record of: No data, band UNKNOWN, and the call answered.](/media/splunk-2.0.0-dashboard-no-data.png)

If the dashboard shows **This account cannot read the Whisper API key**, the account needs `list_storage_passwords`; see [Requirements](/docs/integrations/splunk/requirements#roles-and-capabilities).

## Scheduled searches

A search you schedule over these commands keeps sending its indicators on that schedule, unattended, under the administrator's acknowledgement. Before you alert on what Whisper knows, keep the rows whose call completed and count the rest, so a failed run shows up instead of looking quiet. [Coverage and call status](/docs/integrations/splunk/coverage-and-status#reading-them-together) shows the pattern.
