Scheduled Job Creation via At Utility Inside a Container


Description

This rule detects execution of the "at" or "atd" job-scheduling utilities inside a Linux container. Containers are typically immutable, single-purpose, and do not run interactive job schedulers, so the use of "at" to queue a one-time command is anomalous. Adversaries abuse "at" to establish persistence or to execute a payload at a later time after gaining a foothold inside a container.

Query · esql

from logs-cloud_defend.process* metadata _id, _index, _version
| where host.os.type == "linux" and event.type == "start" and event.action == "exec" and
    container.id like "*" and
    (
      process.name in ("at", "atd") or
      kql("""process.args: "at" """)
    )
| keep @timestamp, host.id, container.id, process.name, process.args, process.parent.name, process.parent.executable, user.id, _id, _index, _version

Known false positives

  • Some container images bundle the "at"/"atd" utilities and a small number of workloads may legitimately schedule one-time jobs. Investigate the queued command and the scheduling principal; baseline and except known automation images where appropriate.

Analyst notes

This rule requires the Elastic Defend for Containers integration to be configured to collect container process events.

Triage and analysis

Investigating Scheduled Job Creation via At Utility Inside a Container

The "at" utility schedules a command to run once at a later time via the "atd" daemon. Inside a container this is rarely legitimate: container workloads are usually immutable and process-bound, with scheduling handled by the orchestrator (CronJobs), not an in-container scheduler. Adversaries use "at" to persist or to defer execution of a payload after compromising a container.

Possible investigation steps

  • Identify the container (container.id, container.name) and namespace (orchestrator.namespace) where "at" executed, and confirm whether in-container scheduling is expected for that workload.
  • Review the parent process (process.parent.name, process.parent.executable) to determine whether the invocation came from an interactive shell or an automated process.
  • Inspect the command that was queued (review the container's /var/spool/cron/atjobs or atq/at -c output) to understand what will run and when.
  • Correlate with prior activity in the same container, such as tool installation, package manager use, or shell access.

False positive analysis

  • A small number of images legitimately run "atd" and may schedule maintenance jobs. Baseline these images or namespaces and add exceptions.

Response and remediation

  • If unexpected, isolate the container, remove the scheduled job (at -r), capture the queued command for analysis, and redeploy the workload from a trusted image.
  • Rotate any credentials exposed to the container and review how the attacker gained execution inside it.
Raw source Scheduled Job Creation via At Utility Inside a Container · Elastic TOML
Esc
Published by elastic/detection-rules ↗, licensed under Elastic License 2.0 ↗. Reproduced here unmodified.
[metadata]
creation_date = "2026/11/13"
integration = ["cloud_defend"]
maturity = "production"
min_stack_comments = "Defend for Containers integration was re-introduced in 9.3.0"
min_stack_version = "9.3.0"
updated_date = "2026/11/13"

[rule]
author = ["Elastic"]
description = """
This rule detects execution of the "at" or "atd" job-scheduling utilities inside a Linux container. Containers are
typically immutable, single-purpose, and do not run interactive job schedulers, so the use of "at" to queue a one-time
command is anomalous. Adversaries abuse "at" to establish persistence or to execute a payload at a later time after
gaining a foothold inside a container.
"""
false_positives = [
    """
    Some container images bundle the "at"/"atd" utilities and a small number of workloads may legitimately schedule
    one-time jobs. Investigate the queued command and the scheduling principal; baseline and except known automation
    images where appropriate.
    """,
]
from = "now-6m"
interval = "5m"
language = "esql"
license = "Elastic License v2"
name = "Scheduled Job Creation via At Utility Inside a Container"
note = """## Setup

This rule requires the Elastic Defend for Containers integration to be configured to collect container process events.

## Triage and analysis

### Investigating Scheduled Job Creation via At Utility Inside a Container

The "at" utility schedules a command to run once at a later time via the "atd" daemon. Inside a container this is rarely legitimate: container workloads are usually immutable and process-bound, with scheduling handled by the orchestrator (CronJobs), not an in-container scheduler. Adversaries use "at" to persist or to defer execution of a payload after compromising a container.

### Possible investigation steps

- Identify the container (`container.id`, `container.name`) and namespace (`orchestrator.namespace`) where "at" executed, and confirm whether in-container scheduling is expected for that workload.
- Review the parent process (`process.parent.name`, `process.parent.executable`) to determine whether the invocation came from an interactive shell or an automated process.
- Inspect the command that was queued (review the container's `/var/spool/cron/atjobs` or `atq`/`at -c` output) to understand what will run and when.
- Correlate with prior activity in the same container, such as tool installation, package manager use, or shell access.

### False positive analysis

- A small number of images legitimately run "atd" and may schedule maintenance jobs. Baseline these images or namespaces and add exceptions.

### Response and remediation

- If unexpected, isolate the container, remove the scheduled job (`at -r`), capture the queued command for analysis, and redeploy the workload from a trusted image.
- Rotate any credentials exposed to the container and review how the attacker gained execution inside it.
"""
references = [
    "https://attack.mitre.org/techniques/T1053/002/",
    "https://github.com/redcanaryco/atomic-red-team/blob/master/atomics/T1053.002/T1053.002.md",
]
risk_score = 21
rule_id = "1a263a5f-8aea-4c42-a121-0defc82e0a2e"
severity = "low"
tags = [
    "Data Source: Elastic Defend for Containers",
    "Domain: Containers",
    "OS: Linux",
    "Use Case: Threat Detection",
    "Tactic: Execution",
    "Tactic: Persistence",
    "Resources: Investigation Guide",
    "Rule Type: ES|QL",
    "Platform: Linux",
    "Platform: Kubernetes",
]
timestamp_override = "event.ingested"
type = "esql"

query = '''
from logs-cloud_defend.process* metadata _id, _index, _version
| where host.os.type == "linux" and event.type == "start" and event.action == "exec" and
    container.id like "*" and
    (
      process.name in ("at", "atd") or
      kql("""process.args: "at" """)
    )
| keep @timestamp, host.id, container.id, process.name, process.args, process.parent.name, process.parent.executable, user.id, _id, _index, _version
'''


[[rule.threat]]
framework = "MITRE ATT&CK"
[[rule.threat.technique]]
id = "T1053"
name = "Scheduled Task/Job"
reference = "https://attack.mitre.org/techniques/T1053/"
[[rule.threat.technique.subtechnique]]
id = "T1053.002"
name = "At"
reference = "https://attack.mitre.org/techniques/T1053/002/"



[rule.threat.tactic]
id = "TA0002"
name = "Execution"
reference = "https://attack.mitre.org/tactics/TA0002/"
[[rule.threat]]
framework = "MITRE ATT&CK"
[[rule.threat.technique]]
id = "T1053"
name = "Scheduled Task/Job"
reference = "https://attack.mitre.org/techniques/T1053/"
[[rule.threat.technique.subtechnique]]
id = "T1053.002"
name = "At"
reference = "https://attack.mitre.org/techniques/T1053/002/"



[rule.threat.tactic]
id = "TA0003"
name = "Persistence"
reference = "https://attack.mitre.org/tactics/TA0003/"

Detection rules belong to the projects that publish them and remain under their own licenses. This site indexes and links to them; it claims no rights in them.