Response Actions

Overview

Actions in LimaCharlie Detection & Response () rules define what happens after a detection is triggered. Common actions include generating reports, tagging sensors, isolating networks, and the frequently used task action, which sends commands to an Endpoint Agent to interrogate or take action on the endpoint. This is useful for tasks like gathering system information or isolating a compromised endpoint. Suppression settings manage repetitive alerts by limiting action frequency, ensuring efficient automation and response workflows.

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

Suppression

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
  suppression:
    max_count: 1
    period: 1h
    is_global: true
    keys:
      - '{{ .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. 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.

Supported Time Period Formats

LimaCharlie supports the following formats for time periods: ns, us (or µs, both are accepted), ms, s, m, h (nanoseconds, microseconds, milliseconds, seconds, minutes, and hours, respectively)

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
  suppression:
    min_count: 3
    max_count: 3
    period: 24h

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 occurred on a single host per period.

Note: Both min_count and max_count must be specified when setting a threshold.

Variable Count

It is also possible to increment a suppression by a value that's not one (1). This is achieved using the count_path parameter, which is a path (like event/record/v) pointing to an integer that should be used to increment the suppression counter.

This is useful for things like billing alerts, where we set a threshold activation (meaning "alert me if above X") where the threshold is reached by increments of billable values.

Here's an example of this:

detect:
    event: billing_record
    op: is
    path: event/record/k
    target: billing
    value: ext-strelka:bytes_scanned

respond:
    - action: report
      name: strelka-bytes-reached
      suppression:
        count_path: event/record/v
        is_global: true
        keys:
          - strelka-bytes-usage
        max_count: 1048576
        min_count: 1048576
        period: 24h

The above will alert (generate a detection in this case) when 1MB (1024 x 1024 x 1) of bytes have been billed by the Strelka Extension based on the bytes_scanned SKU, per 24h.

It does so by incrementing the suppression counter by the billed value (found in event/record/v), resetting after 24h, and if the value of 1MB is reached, alert once and only once.

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.

seal

Seals the sensor in a persistent fashion (if the sensor/host reboots, it will remain sealed). Only works on platforms supporting the seal sensor command.

- action: seal

Sealing a sensor enables tamper resistance, preventing direct modifications to the installed EDR.

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

unseal

Removes the seal status of a sensor that had it set using seal.

- action: unseal

output

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.

Example:

- action: output
  name: my-output

rejoin network

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

- action: rejoin network

report

- action: report
  name: my-detection-name
  publish: true # defaults to true
  priority: 3   # optional
  metadata:     # optional & free-form
    author: Alice (alice@wonderland.com)
  detect_data:  # additional free-form field that can be used for extraction of specific elements

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, metadata and detect_data parameters support string templates like detected {{ .cat }} on {{ .routing.hostname }}, note that the context of the transform is the detection itself and not the original event, so you would refer to .detect.event.USER_NAME and not .event.USER_NAME for example.

The metadata is generally used to populate information about the rule, its author, remediation etc.

The detect_data is generally used to extract specific parts of the detected event into a known format that can be common across multiple detection, like extracting the domain or hash field for example.

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.

task

- 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 Endpoint Agent Commands

undelete sensor

Un-deletes a sensor that was previously deleted.

detect:
    target: deployment
    event: deleted_sensor
    op: is
    path: routing/event_type
    value: deleted_sensor
respond:
    - action: undelete sensor

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

wait

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 milliseconds, as defined by this function call.

• An integer representing a number of seconds.

Example:

- action: wait
  duration: 10s

and

- action: wait
  duration: 5

add hive tag

Adds a tag to a Hive record. This can be used to mark some Hive records like D&R rules automatically.

- action: add hive tag
  hive name: dr-general
  record name: my-rule
  tag: high-hit

Unless the rule is not expected to hit often, you likely want to couple this with a suppression statement to avoid doing a lot of tagging of the same rules like:

- action: add hive tag
  hive name: dr-general
  record name: my-rule
  tag: high-hit
  suppression:
    max_count: 1
    period: 1h
    is_global: true
    keys:
      - 'high-hit'
      - 'hive-tag'

remove hive tag

Removes a tag from a Hive record.

- action: remove hive tag
  hive name: dr-general
  record name: my-rule
  tag: high-hit