Skip to main content

Risk-based alerting on Axiom

Most single detections are too weak to act on alone. Several of them on the same user or host in one day usually are. With risk-based alerting, Huntbase gives every insight a risk score, writes that risk to an Axiom dataset for each user, host and IP involved, and adds it up per entity. When an entity's risk crosses a threshold, you get one insight about the entity, linked to everything that added to it.

The risk events stay in your Axiom organisation, next to your other data, so you can query them in APL too.

Turn it on​

  1. In Axiom, create a dataset for risk events, for example huntbase-risk.
  2. Make sure the connection's API token has ingest and query permission on that dataset.
  3. On the Axiom connection in Huntbase, set Send risk events to to the dataset's name, and save.

From then on, every new insight in your organisation writes risk events to that dataset. Leave Send risk events to empty to turn it off.

What Huntbase writes​

For each insight, Huntbase writes one risk event per user, host and IP the insight involves, up to 10. Each event holds:

FieldWhat it holds
_timeWhen the risk event was written
entity_typeuser, host or ip
entityThe user name, host name or address
risk_scoreThe insight's risk, from 1 to 100
ruleThe rule or watcher that raised the insight
mitre_tactics, mitre_techniquesThe insight's ATT&CK tactics and techniques
insight_id, watcher_idLinks back to Huntbase
severity, insight_timeThe insight's severity and when it happened
org_id, sourceYour organisation, and what wrote the event

The risk score comes from the insight's severity and confidence:

SeverityBase score
Informational5
Low10
Medium25
High50
Critical80

The base is multiplied by the insight's confidence: high counts in full, medium three quarters, low half. Two high-confidence High insights on one entity add up to 100.

When it raises an insight​

Every 15 minutes, Huntbase sums each entity's risk events over the last 24 hours. It raises Risk threshold exceeded for <entity> when either:

  • the entity's risk adds up to at least the Risk threshold (100 unless you change it), or
  • its risk events span at least the Tactics threshold of different ATT&CK tactics (3 unless you change it), whatever the score.

Either way, the risk must come from at least Detections needed separate insights (2 unless you change it), raised by different rules. One noisy rule, or one rule tagged with several tactics, can't raise it alone. The exception is a single insight whose own risk reaches the threshold.

All three settings are under Advanced on the connection.

The insight is High, or Critical when the risk reaches twice the threshold. It sits on the user, host or IP, so it shows on that entity's page. Huntbase raises it at most once per entity per day.

The insight's Risk roll-up section shows the entity's risk against the threshold, for example Risk 140 in 24 hours (threshold 100), why it fired, the ATT&CK tactics and rules involved, and links to the insights that added to it, by title.

Risk roll-up insights don't write risk events themselves, so they never add to their own score.

Query risk in Axiom​

The Axiom connection's query templates include Risk roll-up (24h), the same sum Huntbase runs, and Risk events for an entity, which lists what added to one user's or host's risk. Find them in Content library › Queries, and set the dataset to your risk dataset.

The roll-up runs one small APL query on the risk dataset every 15 minutes, which uses a little Axiom query compute. See Keep query compute low.

Next steps​