Detection rules › Splunk

Excessive File Deletion In WinDefender Folder

Author
Teoderick Contreras, Splunk, Steven Dick
Source
upstream

The following analytic detects excessive file deletion events in the Windows Defender folder. It leverages Sysmon EventCodes 23 and 26 to identify processes deleting multiple files within this directory. This behavior is significant as it may indicate an attempt to corrupt or disable Windows Defender, a key security component. If confirmed malicious, this activity could allow an attacker to disable endpoint protection, facilitating further malicious actions without detection.

MITRE ATT&CK coverage

TacticTechniques
ImpactT1485 Data Destruction

Event coverage

ProviderEvent IDTitle
Sysmon23FileDelete (File Delete archived)
Sysmon26FileDeleteDetected (File Delete logged)

Stages and Predicates

Stage 1: search

search EventCode IN ("23", "26") TargetFilename="*\\ProgramData\\Microsoft\\Windows Defender\\*"

Stage 2: stats

stats BY action, dest, dvc, signature, signature_id, user, user_id, vendor_product

Stage 3: where

where count>=50

Stage 4: search

search

Stage 5: search

search

Stage 6: search

search `macro`

Indicators

Each row is a field, operator, and value that the rule matches. The corpus column counts how many other rules in the catalog look for the same combination: high numbers point to widely-used, community-vetted indicators. Blank or 1 shows that the indicator is specific to this rule.

FieldKindValues
EventCodein
  • "23" corpus 6 (splunk 6)
  • "26" corpus 6 (splunk 6)
TargetFilenameeq
  • "*\\ProgramData\\Microsoft\\Windows Defender\\*"
countge
  • 50

Neighbors

Often fire together

Rules that target events appearing in the same incident timelines. They pattern-match on adjacent steps of the same TTP, so an alert from one is often paired with alerts from these. Useful for triage context and for assembling chained-detection rules.

Share event IDs (chain-detection candidates)

Rules that observe the same Windows event-ID pairs as this one. If you're authoring a multi-stage / sequence rule that spans these events, these are the existing detections that already cover one or both endpoints.