
Summary
Detects execution of the at/atd job-scheduling utilities inside Linux containers. Containers are typically immutable, single-purpose environments where in-container schedulers are unusual; attackers may abuse at to queue commands for persistence or deferred execution after compromising a container. The rule flags Linux container start events where a process named at or atd is invoked, or where the command line contains an at invocation, leveraging Elastic Defend for Containers process events. The accompanying ES|QL query extracts key fields such as container and host identifiers, process details, and parent process context to aid investigation. Techniques mapped include MITRE ATT&CK T1053.002 (At) under Execution and T1053 (Scheduled Task/Job) under Persistence. The rule supports cross-reference with at-spool or atq/at -c outputs to understand what will run and when, and what user contexts invoked the scheduling. The rule is designed to help responders determine whether in-container scheduling is legitimate or indicative of adversary activity.
### Investigation focus
- Identify the affected container and namespace (container.id, container.name, orchestrator.namespace).
- Review the parent process to determine if the invocation came from an interactive shell or automated process.
- Inspect the scheduled job details (queued command, timing) via in-container cron spool or atq/at -c outputs.
- Correlate with prior container activity (tool installs, shell access, credential usage).
### False positives
- Some images legitimately run atd to perform maintenance tasks. Baseline such images/namespaces and apply exceptions as appropriate.
### Remediation
- If suspicious, isolate the container, remove the scheduled job (at -r), capture the queued command for analysis, and redeploy from a trusted image.
- Rotate credentials exposed to the container and review attack chain to understand how access was gained.
Categories
- Containers
Data Sources
- Process
ATT&CK Techniques
- T1053
- T1053.002
Created: 2026-11-13