Reference: Actions
  • 07 Feb 2024
  • 7 Minutes to read
  • Contributors
  • Dark

Reference: Actions

  • Dark

Article Summary

Actions are used in the Response part of a Detection & Response rule.

For more information on how to use Actions, read Detection & Response rules.


Suppression is valuable to help manage repetitive or noisy alerts.

Reduce Frequency

In some cases, you may want to limit the number of times a specific Action is executed over a certain period of time. You can achieve this through suppression. This feature is supported in every Actions.

A suppression descriptor can be added to an Action like:

- action: report
  name: evil-process-detected
    max_count: 1
    period: 1h
    is_global: true
      - '{{ .event.FILE_PATH }}'
      - 'evil-process-detected'

The above example means that the evil-process-detected detection will be generated up to once per hour per FILE_PATH. Beyond the first report with a given FILE_PATH, during the one hour period, new report actions from this rule will be skipped.

The is_global: true means that the suppression should operate globally within the Org (tenant), if the value was false, the suppression would be scoped per-sensor.

The keys parameter is a list of strings that support [templating](/v1/docs/platform-management-template strings). Together, the unique combination of values of all those strings (ANDed) will be the uniqueness key this suppression rule uses. By adding to the keys the {{ .event.FILE_PATH }} template, we indicate that the FILE_PATH of the event generating this report is part of the key, while the constant string evil-process detected is just a convenient way for us to specify a value related to this specific detection. If the evil process-detected component of the key was not specified, then all actions that also just specify the {{ .event.FILE_PATH }} would be contained in this suppression. This means that using is_global: true and a complex key set, it is possible to suppress some actions across multiple Actions across multiple D&R rules.

Threshold Activation

The other way to use suppression is using the min_count parameter. When set, the specific action will be suppressed until min_count number of activations have been received in that period.

Here's an example of this:

- action: report
  name: high-alerts
    min_count: 3
    max_count: 3

The above example means the high-alerts detection will be generated once per hour but only after the rule the action belongs to has matched 3 times within that period.

This could be useful if you wanted to create higher order alerts that trigger a different type of detection, or send a page alert to a SOC, when more than X alerts occured on a single host per period.

Available Actions

Actions allow you to specify "what" happens after a detection is found.

add tag, remove tag

- action: add tag
  tag: vip
  entire_device: false # defaults to false
  ttl: 30 # optional

Adds or removes tags on the sensor.

Optional Parameters

The add tag action can optionally take a ttl parameter that is a number of seconds the tag should remain applied to the sensor.

The add tag action can optionally have the entire_device parameter set to true. When enabled, the new tag will apply to the entire Device ID, meaning that every sensor that shares this Device ID will have the tag applied (and relevant TTL). If a Device ID is unavailable for the sensor, it will still be tagged.

This can be used as a mechanism to synchronize and operate changes across an entire device. A D&R rule could detect a behavior and then tag all sensors on the device so they may act accordingly, like start doing full pcap.

For example, this would apply the full_pcap to all sensors on the device for 5 minutes:

- action: add tag
  tag: full_pcap
  ttl: 300
  entire_device: true

add var, del var

Add or remove a value from the variables associated with a sensor.

- action: add var
  name: my-variable
  value: <<event/VOLUME_PATH>>
  ttl: 30 # optional

The add var action can optionally take a ttl parameter that is a number of seconds the variable should remain in state for the sensor.

extension request

Perform an asynchronous request to an extension the organization is subscribed to.

- action: extension request
  extension name: dumper # name of the extension
  extension action: dump # action to trigger
  extension request:     # request parameters
    sid: '{{ .routing.sid }}'
    pid: event.PROCESS_ID

The extension request parameters will vary depending on the extension (see the relevant extension's schema). The extension request parameter is a transform.

You can also specify a based on report: true parameter. When true (defaults to false), the transform for the extension request will be based on the latest report action's report instead of the original event. This means you MUST have a report action before the extension request.

isolate network

Isolates the sensor from the network in a persistent fashion (if the sensor/host reboots, it will remain isolated). Only works on platforms supporting the segregate_network sensor command.

- action: isolate network

When the network isolation feature is used, LimaCharlie will block connections to all destinations other than the LimaCharlie cloud (so that you can perform an investigation, take remediation actions, and then ultimately remove the isolation to resume normal network operation). The host will maintain internet connectivity to allow for you to perform those actions.

The segregate_network command is stateless, so if the endpoint reboots, it will not be in effect. The isolate network command in D&R rules is stateful, so it sets a flag in the cloud to make sure the endpoint remains isolated even after reboots.


Forwards the matched event to an Output identified by name in the tailored stream.

This allows you to create highly granular Outputs for specific events.

The name parameter is the name of the Output.


- action: output
  name: my-output

rejoin network

Removes the isolation status of a sensor that had it set using isolate network.

- action: rejoin network


- action: report
  name: my-detection-name
  publish: true # defaults to true
  priority: 3   # optional
  metadata:     # optional & free-form
    author: Alice (

Reports the match as a detection. Think of it as an alert. Detections go a few places:

  • The detection Output stream
  • The organization's Detections page (if insight is enabled)
  • The D&R rule engine, for chaining detections

The name and metadata parameters support string templates like detected {{ .cat }} on {{ .routing.hostname }}.

Limiting Scope

There is a mechanism for limiting scope of a report, prefixing name with __ (double underscore). This will cause the detection
generated to be visible to chained D&R rules and Services, but the detection will not be sent to the Outputs for storage.

This is a useful mechanism to automate behavior using D&R rules without generating extra traffic that is not useful.

Optional Parameters

The priority parameter, if set, should be an integer. It will be added to the root of the detection report as priority.

The metadata parameter, if set, can include any data. It will be added to the root of the detection report as detect_mtd. This can be used to include information for internal use like reference numbers or URLs.

service request

Perform an asynchronous request to a service the organization is subscribed to.

- action: service request
  name: dumper # name of the service
  request:     # request parameters
    sid: <<routing/sid>>
    retention: 3

The request parameters will vary depending on the service (see the relevant service's documentation). Parameters can also leverage lookback values (i.e. <<path/to/value>>) from the detected event. The request parameter also support string templates.

You can also specify a based on report: true parameter. When true (defaults to false), the lookback and template strings for the request will be based on the latest report action's report instead of the original event. This means you MUST have a report action before the service request.


- action: task
  command: history_dump
  investigation: susp-process-inv

Sends a task in the command parameter to the sensor that the event under evaluation is from.

An optional investigation parameter can be given to create a unique identifier for the task and any events emitted from the sensor as a result of the task.

The command parameter supports string templates like artifact_get {{ .event.FILE_PATH }}.

To view all possible commands, see Reference: Sensor Commands

undelete sensor

Un-deletes a sensor that was previously deleted.

    event: DELETED_SENSOR      
    op: is      
    path: routing/event_type      
    value: DELETED_SENSOR    
- action: undelete sensor

This can be used in conjunction with the deleted_sensor event to allow sensors to rejoin the fleet.


Adds a delay (up to 1 minute) before running the next response action.

This can be useful if a previous response action needs to finish running (i.e. a command or payload run via task) before you can execute the next action.

The wait action will block processing any events from that sensor for the specified duration of time. This is because D&R rules are run at wire-speed and in-order.

The duration parameter supports two types of values:

  • A string describing a duration, like 5s for 5 seconds or 10ms for 10 miliseconds, as defined by this function call.
  • An integer representing a number of seconds.


- action: wait
  duration: 10s


- action: wait
  duration: 5

Was this article helpful?