Modifying Detections with Inline Filters (Beta)

Modify an existing rule without writing code

Overview

Inline filters are in open beta as of Panther version 1.54. Please share any bug reports and feature requests with your account team.

You can easily tune existing rules, including Panther-managed rules, by adding Inline Filters. An Inline Filter is a condition that must pass in order for the detection logic to then be run. Inline Filters are available only on rules, not scheduled rules nor policies.

In the Panther Console, you can create Inline Filters using a no-code builder. In the CLI workflow, you can create Inline Filters by adding the InlineFilters YAML key.

A common use case for filters is to add an allowlist or denylist.

How Inline Filters work

Filter statements are evaluated before a detection's logic. A filter must return true (i.e., match the event) for the detection logic itself to then be run.

In both the Console and CLI workflow, filters can be grouped using AND or OR logic.

If an event does not contain the field the filter is evaluating, the filter will pass. If the field the filter is evaluating has a value of none, the filter will return false on positive comparators or on comparators that don't apply, and true for inverse comparators.

In the Console, filters are not available during new rule creation. In the CLI workflow, you can include InlineFilters on new rules.

While it is broadly discouraged to manage detection content using both CLI workflows and the Console simultaneously, it is possible to use Inline Filters in the Console alongside the CLI workflow. Filters created in the Console will not be overwritten or deleted when an update to detection content is made in the CLI workflow.

Creating filters in the Panther Console

You can add filters to a rule from its edit page, or within an alert triggered by that rule.

From a rule's edit page

Add filters from a rule's edit page

1. In the left-hand navigation bar of your Panther Console, click Detections.

2. In the list of detections, click a rule's name to view its details page.

3. Within the Detect section, under Filter to only include events: and to the right of Where, click +.

   • In the menu that appears, select either Add Filter or Add Filter Group.

4. For each filter (either on its own or within a group), define the logic:

   1. Click Key, then select an event key the condition will apply to.

      • To indicate a nested field, use JSON path notation.

      • Some options may include [*], indicating the key is an array of objects. Learn more about indexing an array of objects below.

   2. Click Condition, then select a condition.

   3. If the selected Condition requires an inputted value(s) (e.g., is or contains), provide a value or list of values.

      • If the values(s) field takes in an array, see the Inputting array values instructions below.

5. Between each filter and filter group, ensure the correct combinator (either and or or) is selected.

6. Run unit tests to ensure they pass with the added filter(s).

7. In the upper-right corner of the page, click Deploy to save your changes.

From an alert event

Add filters from an alert event

You can add Inline Filters to a rule directly from an event in an associated alert. This is particularly helpful if you've received a false positive alert, and want to tune the triggered detection so it won't match on similar events in the future.

1. In the left-hand navigation bar of your Panther Console, click Alerts.

2. Locate the alert whose associated rule you'd like to tune, and click its name.

3. On the alert's detail page, scroll down to the Event section.

4. In the event's JSON, hover over the indicator you'd like the new filter to target, and click the target icon.

   • The Add Filter slide-out panel will open on the right-hand side of the window.

5. In the Add Filter slide-out panel, a new filter will be pre-populated in the following way:

   • Key: defaults to the field on which you clicked the target icon in the event JSON.

   • Condition: defaults to is not, assuming you would not like to receive alerts for events like this in the future.

   • String: defaults to the value of the selected field in the event JSON.

6. Make any desired changes to the filter. All pre-populated fields (i.e., Key, Condition and String) are editable.

7. Locate the Unit Test section near the bottom of the panel. If the rule is not Panther-managed and you'd like to create a new unit test for the rule using the current event, click the checkbox labeled Add the current alert event as a unit test.

   • The toggle labeled The detection should trigger based on the example event is editable. It defaults to No, as you are likely trying to prevent alerts like this in the future.

   • If the rule is Panther-managed, this option will be greyed out.

8. Click Save & Run Test.

   • This runs all of the target rule's unit tests. If you created a new unit test in step 7, it is also run.

   • In order for the new filter(s) to be saved, all of the rule's unit tests must pass. If any of the unit tests fail:

     • If the rule is not Panther-managed, click View Detection to be taken to the rule's detail page to edit unit tests. From there, you can click Update to save your changes to the rule.

     • If the rule is Panther-managed, its unit tests are read-only, meaning you can't alter failing tests to make them pass. To be able to add the filter successfully, instead follow the Working with failed unit tests with filters workflow.

Inputting array values

If the Rule Filter operator you've selected requires the value field to take in an array (such as the is in operator), you'll input the array values in a modal that pops up when you click into the value field.

To add values to an array:

1. After selecting a Key and Condition for your Filter, click into the values field.

   

   • This will open the array input modal.

2. In the modal, enter the array value(s) in the input field.

   • If your input is comma-delimited, check the Values entered above are comma-delimited checkbox.

     • When this field is checked, the text inputted into the values field will be separated (using a comma delimiter) into multiple values. For example, entering "User 1,User 2,User 3" will result in three values added.

       
   • If your input is not comma-delimited, leave Values entered above are comma-delimited unchecked.

     • When this field is unchecked, you can add values that contain commas one at a time. For example, entering "1,000" will add just one value.
3. Click Add.

4. Repeat steps 2-3 as needed, until all values have been added to the array.

5. Click Apply.

Indexing an array of objects

While creating a filter expression, if there is an event key whose value is an array of objects, that key will be shown in the dropdown selector with the array indexing symbol [*], along with the fields in the object. You can use [*] to target the selected field in all objects in the array, or replace * with an integer to index the array, targeting a single field.

Example

For example, take this resources field in the AWS.CloudTrail schema:

It will be represented in the event field selector as:

By default, Panther applies a wildcard array index ([*]) that will search across values for the chosen field in all objects in the array. When [*] is used, an array of these values is created and searched. Because of this, only the array conditions are available: is empty, is not empty, contains, does not contain.

For example, the below filter expression (resources[*].type contains AWS::IAM::Role) means an event will match if any value of type in the resources array is AWS::IAM::Role.

However, you can replace the * with an integer to index the array, which specifies a single object in the array. In this case, Panther will only evaluate the value of the nested field at that index.

The conditions shown will be updated to those which are applicable to the data type of the nested field chosen:

Creating filters in the CLI workflow

In addition to creating no-code rule filters in the Panther Console, you can also create YAML filters on your rules written as Simple Detections or Python detections.

Like the filters created in the Console, YAML filters are evaluated before the detection logic of a rule. If the filter returns true, the detection logic will be executed. If the filter returns false, the evaluation of the detection will stop, and the detection will return false altogether.

YAML InlineFilter syntax

A YAML filter is denoted by the InlineFilters key. Within InlineFilters, list one or more match expressions. You can use the All and Any combinators to specify AND or OR logic, respectively, and nest combinators to create filter groups. If a combinator is not specified directly under InlineFilters, All is assumed.

See Simple Detection Match Expression Reference to learn how to construct different types of match expressions.

Example:

InlineFilters: 
  - KeyPath: environment
    Condition: StartsWith
    Value: "Sandbox"

Limitations of YAML Inline Filters

Some match expression functionality described in Simple Detection Match Expression Reference is not possible in InlineFilters. These limitations include:

• InlineFilters cannot be used on scheduled rules or policies, only rules.

• The following match expression types cannot be used within InlineFilters:

  • Multi-key match expressions
  • List comprehension match expressions
  • Absolute match expressions
  • Enrichment match expressions
• The Key and DeepKey key specifiers cannot be used within InlineFilters—only KeyPath may be used.
• The OnlyOne and None combinators cannot be used within InlineFilters—only All and Any may be used.
• Many Condition values cannot be used within InlineFilters. Only the following values may be used:

  • Equals

  • DoesNotEqual

  • IsGreaterThan

  • IsGreaterThanOrEquals

  • IsLessThan

  • IsLessThanOrEquals

  • Contains

  • DoesNotContain

  • StartsWith

  • EndsWith

  • IsIPAddressInCIDR

  • IsIPAddressNotInCIDR

  • CIDRContainsIPAddresses

  • CIDRDoesNotContainIPAddresses

  • IsIn

  • IsNotIn

  • IsIPAddressPublic

  • IsIPAddressPrivate

  • IsNullOrEmpty

  • IsNotNullOrEmpty

How to create an Inline Filter in the CLI workflow

To create an Inline Filter in the CLI workflow on a rule created as either a Python or Simple Detection, in the detection's YAML file, include the InlineFilters key. Within InlineFilters, include one or more match expressions.

Working with failed unit tests with filters

For Panther-managed rules with filters, you currently cannot add or edit unit tests. You cannot save a rule if the unit test does not pass.

If a unit test fails, take the following steps:

1. Clone the Panther-managed rule.

2. Add your filter(s) to the cloned rule.

3. Edit the unit tests for the cloned rule so that they pass.

Inline filter reference

Refer to the below operators and value types when building out your filters in the Console.

Supported Console operators

Operation	Usage guidelines	Supported field types	Examples
is / is not	An event will match when the field matches/does not match the value in the filter	string, ip, bool, int	username is “root”
is in / is not in	An event will match when the field matches/does not match an entry in the list of values in the filter	string, int	username is in [ “root”, “admin” ] port is in [25, 553]
is empty	An event will match when the field's value is not specified. The operator tests only for the absence of data	string, int array, ip array, float array, bool array, string array	errors_list is empty
is not empty	An event will match when the field's value is specified. The operator tests only for the presence of data	string, int array, ip array, float array, bool array, string array	errors_list is not empty
contains	An event will match when the value of the field specified contains the value provided When the event value is a string or a string array, partial matching is supported	string, int array, ip array, bool array, string array	domain contains “.google.com” p_any_port contains 22
does not contain	An event will match when the value of the field specified does not contain the value provided When the event value is a string or a string array, partial matching is supported	string, int array, ip array, bool array, string array	domain !contains “.google.com” p_any_port !contains 22
starts with	An event will match when the value of the field specified begins with the value provided	string	role starts with “admin_”
ends with	An event will match when the value of the field specified ends with the value provided	string	domain ends with “.cc”
is greater than	An event will match when the field's value is greater than the value provided in the filter	int, float	port > 1023
is less than	An event will match when the field's value is less than the value provided in the filter	int, float	port < 1024
is greater than or equal	An event will match when the field's value is greater than or equal to the value provided in the filter	int	count ≥ 1
is less than or equal	An event will match when the field's value is less than or equal to the value provided in the filter	int	count ≤ 100
is private	An event will match when the IP address specified is private	IP	dst_ip is_private
is public	An event will match when the IP address specified is public	IP	src_ip is_public
is in CIDR / is not in CIDR	An event will match when the IP address specified is/is not within a provided CIDR (Classless Inter-Domain Routing) block	IP	src_ip in_cidr 192.168.0.0/16
does not contain IP in CIDR	An event will match when the IP array specified does not contain any IP address within the CIDR block provided	ip array	p_any_ip_address !contains_ip 8.8.0.0/16 p_any_ip_address !contains_ip 1.1.1.1/32
contains IP in CIDR	An event will match when the IP array specified contains an IP address within the CIDR block provided	ip array	p_any_ip_address contains_ip 8.8.0.0/16 p_any_ip_address contains_ip 1.1.1.1/32

Supported value types

Value types	Description
string	A string value
int	A 32-bit integer number in the range -2147483648, 2147483647
float	A 64-bit floating point number
boolean	A boolean value true / false
array	A JSON array where each element is of the same type
ip	A single valid IPv4 or IPv6 address
CIDR	A classless inter-domain routing block