Unusually Long Command Line - MLTK

SPLUNK detection rule by Rico Valdez, Splunk · Apache-2.0

The following analytic identifies unusually long command lines executed on hosts, which may indicate malicious activity. It leverages the Machine Learning Toolkit (MLTK) to detect command lines with lengths that deviate from the norm for a given user. This is significant for a SOC as unusually long command lines can be a sign of obfuscation or complex malicious scripts. If confirmed malicious, this activity could allow attackers to execute sophisticated commands, potentially leading to unauthorized access, data exfiltration, or further compromise of the system.

name: Unusually Long Command Line - MLTK
id: 57edaefa-a73b-45e5-bbae-f39c1473f941
version: 9
creation_date: '2020-04-29'
modification_date: '2026-05-13'
author: Rico Valdez, Splunk
status: removed
type: Anomaly
description: The following analytic identifies unusually long command lines executed on hosts, which may indicate malicious activity. It leverages the Machine Learning Toolkit (MLTK) to detect command lines with lengths that deviate from the norm for a given user. This is significant for a SOC as unusually long command lines can be a sign of obfuscation or complex malicious scripts. If confirmed malicious, this activity could allow attackers to execute sophisticated commands, potentially leading to unauthorized access, data exfiltration, or further compromise of the system.
data_source:
    - Sysmon EventID 1
    - Windows Event Log Security 4688
    - CrowdStrike ProcessRollup2
search: |-
    | tstats `security_content_summariesonly` count min(_time) as firstTime max(_time) as lastTime FROM datamodel=Endpoint.Processes
      BY Processes.action Processes.dest Processes.original_file_name
         Processes.parent_process Processes.parent_process_exec Processes.parent_process_guid
         Processes.parent_process_id Processes.parent_process_name Processes.parent_process_path
         Processes.process Processes.process_exec Processes.process_guid
         Processes.process_hash Processes.process_id Processes.process_integrity_level
         Processes.process_name Processes.process_path Processes.user
         Processes.user_id Processes.vendor_product
    | `drop_dm_object_name(Processes)`
    | `security_content_ctime(firstTime)`
    | `security_content_ctime(lastTime)`
    | eval processlen=len(process)
    | search user!=unknown
    | apply cmdline_pdfmodel threshold=0.01
    | rename "IsOutlier(processlen)" as isOutlier
    | search isOutlier > 0
    | table firstTime lastTime user dest process_name process processlen count
    | `unusually_long_command_line___mltk_filter`
how_to_implement: The detection is based on data that originates from Endpoint Detection and Response (EDR) agents. These agents are designed to provide security-related telemetry from the endpoints where the agent is installed. To implement this search, you must ingest logs that contain the process GUID, process name, and parent process. Additionally, you must ingest complete command-line executions. These logs must be processed using the appropriate Splunk Technology Add-ons that are specific to the EDR product. The logs must also be mapped to the `Processes` node of the `Endpoint` data model. Use the Splunk Common Information Model (CIM) to normalize the field names and speed up the data modeling process. In addition, the Machine Learning Toolkit (MLTK) version 4.2 or greater must be installed on your search heads, along with any required dependencies. Finally, the support search "ESCU - Baseline of Command Line Length - MLTK" must be executed before this detection search, because it builds a machine-learning (ML) model over the historical data used by this search. It is important that this search is run in the same app context as the associated support search, so that the model created by the support search is available for use. You should periodically re-run the support search to rebuild the model with the latest data available in your environment.
known_false_positives: Some legitimate applications use long command lines for installs or updates. You should review identified command lines for legitimacy. You may modify the first part of the search to omit legitimate command lines from consideration. If you are seeing more results than desired, you may consider changing the value of threshold in the search to a smaller value. You should also periodically re-run the support search to re-build the ML model on the latest data. You may get unexpected results if the user identified in the results is not present in the data used to build the associated model.
references: []
rba:
    message: Unusually long command line usage on $dest$
    risk_objects:
        - field: user
          type: user
          score: 20
        - field: dest
          type: system
          score: 20
    threat_objects: []
tags:
    analytic_story:
        - Suspicious Command-Line Executions
        - Unusual Processes
        - Possible Backdoor Activity Associated With MUDCARP Espionage Campaigns
        - Ransomware
    asset_type: Endpoint
    product:
        - Splunk Enterprise
        - Splunk Enterprise Security
        - Splunk Cloud
    security_domain: endpoint
deprecation_info:
    reason: Detection is deprecated as these do not work with the latest Splunk AI Toolkit(5.7.0) and Python for Scientific Computing for Linux 64-bit(4.3.0).
    removed_in_version: 5.26.0
    replacement_content: []

[[ currentRule.title && currentRule.title.length > 60 ? currentRule.title.slice(0, 60) + '…' : currentRule.title ]]

7348e846-a52f-442f-ab2a-aba326f7ba33

Propose edits, view related rules, and engage with the community through comments.

Executable code, not a declarative rule

This submission is arbitrary [[ currentRule.format ? currentRule.format.toUpperCase() : '' ]] code, not a declarative detection rule — review it as you would any script before running it, and note that it cannot be safely aggregated or auto-executed alongside YARA/Sigma/Suricata/CRS rules.

Rejected pattern present

[[ reason ]]

What does this mean?

Format & documentation suggestions

Not a security issue — this rule is valid and was accepted as-is. These are optional fields/conventions its own format recommends, mainly affecting its quality score and how well it documents itself for other reviewers.

[[ reason ]]

What do these messages mean?

[[ currentRule.title && currentRule.title.length > 60 ? currentRule.title.slice(0, 60) + '…' : currentRule.title ]]

[[ currentRule.title ]]

Community Discussion