Tools · SIEM and detection engineering · 18 / 24
Validating detections with Atomic Red Team in a lab
Use small, documented Atomic Red Team tests in an isolated lab VM to confirm that telemetry is collected and that your rules fire, then clean up and record the results.
What atomics are
Atomic Red Team is an open-source library from Red Canary of small tests mapped to ATT&CK techniques. Each atomic is a short, documented action, such as creating a scheduled task or running a discovery command, with a cleanup step. They are meant for defenders to check whether their telemetry and detections work, and they belong only on lab or explicitly approved systems.
Run safely
Use an isolated, snapshotted VM that you own, with your endpoint agent and log forwarding installed. Read each test’s YAML before running it, prefer tests marked as safe discovery or persistence simulations, and always run the cleanup command. Get written approval before running any test outside a personal lab.
- Isolated VM with a snapshot
- Read the atomic YAML first
- Run the cleanup
- Approval for anything beyond your lab
Lab: test a scheduled task detection
On a Windows lab VM, install the Invoke-AtomicRedTeam PowerShell module, list the details of the scheduled task technique, check prerequisites, run one test, confirm in your SIEM that Event ID 4698 and the Sysmon process creation for schtasks.exe arrived and your rule fired, then run cleanup and revert the snapshot.
IEX (IWR 'https://raw.githubusercontent.com/redcanaryco/invoke-atomicredteam/master/install-atomicredteam.ps1' -UseBasicParsing); Install-AtomicRedTeam -getAtomics Invoke-AtomicTest T1053.005 -ShowDetailsBrief Invoke-AtomicTest T1053.005 -TestNumbers 1 -CheckPrereqs Invoke-AtomicTest T1053.005 -TestNumbers 1 # check SIEM for 4698 and the rule alert, then: Invoke-AtomicTest T1053.005 -TestNumbers 1 -Cleanup
Record results
For each test, record whether telemetry arrived, whether the rule fired, the delay, and any gaps. A missing alert can mean a rule bug, a parsing issue, or missing data, and each needs a different fix. Repeat the tests after SIEM or agent upgrades, because pipelines break silently.
Diagram
Possible outcomes of a validation test
Each points to a different fix.
- Telemetry and alert
Working. Record the delay.
- Telemetry, no alert
Rule logic or field mapping problem.
- No telemetry
Collection gap: audit policy, agent, or forwarding.
- Alert, wrong context
Rule fires but lacks fields analysts need.
Check yourself
1. What is Atomic Red Team intended for?
2. A test ran but no events reached the SIEM. What does that indicate?
3. What should you always do after running an atomic test in the lab?