Salem Cyber Doc Site
  • 🏠Documentation Home
  • ✨Initiation Guides
    • Quickstart: Deploy Salem
    • Admin Guide
    • Installing Teams App
    • Installing Browser Extension
    • Feature Overview
  • ✨General Guides
    • Managing Alerts
    • Managing Questions
    • Threat Notification Management
    • Uploading Files
    • Logical Operations
  • ✨Configurations Specification
    • Configuration Home
    • Action Conf
      • "match" ActionConfs
      • "webhook" ActionConfs
      • "llm" ActionConfs
    • Action Definition
      • Azure Log Analytics
      • Microsoft Graph API
      • Splunk Search
      • Bring Your Own LLM
    • Parsing Conf
      • Summary Details
    • Report Conf
    • LLM Configuration
  • 💾Changelog
    • Dec 5th '24: Get cracking on your holiday shopping list
    • July 18th, ’24: Beat the heat and the hackers
    • Apr 17th, '24: Alert showers make analysts sour... no longer with Salem!
    • Mar 5, '24: They're after me (and your) secure systems! We're na-tur-ally suspicious
    • Jan 31, '24: New year, new me... and a new way to extract data from your alerts
    • Dec 21, '23: Jingle bells, WannaCry smells, your escalated alert just laid an egg
    • Nov 14, '23: Stuff the turkey or stuff cyber alerts with context... Why not both?
    • Oct 25, '23: Llama, llama on the wall which alert is scariest of them all
    • Sept 19, '23: Context building via true positive/false positive workflow
    • Sept 1, '23: Alert report UI, webhook actions, and question upgrades
Powered by GitBook
On this page
  1. Configurations Specification
  2. Action Conf

"match" ActionConfs

PreviousAction ConfNext"webhook" ActionConfs

Last updated 11 months ago

Match actions are created by users when teaching Salem how something was known and can be modified within the Admin menu by entering view -a in Salem Chat and then selecting Configs > ActionConfs.

Below are a few examples of how Salem creates and manages match actions.

Situation: The user answers a multiContext question indicating the type of authorization an action represents. To build Salem's learning, the user wants to provide

  1. The user selects an alert's multiContext question about whether the action related to the alert was authorized. Note: multiContext questions can be answered through the "Add Context" or "Answer a Question" workflow.

  1. The user selects "authorized" based on insight into the action of the selected alert.

  1. The user selects "Evaluate Alert Detail" and enters result == ‘success’ and bytes_out > 5000 and match(‘endsWith', ‘salemcyber.com’, account), and selects "Submit."

  1. Salem generates an ActionConf object which includes, among other components, the following paramaters:

“returns”: [
         “action.authorized”
]
“params”: {
         result == ‘success’ and 
         bytes_out > 5000 and 
         match(‘endsWith', ‘salemcyber.com’, account)
}
  1. Now Salem has learned to identify the actions of alerts as "authorized" if they have the characteristics input by the user.

For a complete list of eval operations, see the documentation.

✨
logical operations
Creating Match Actions with multiContext Questions