Unusual Process Resolving AWS ECS Agent Communication Service Endpoint

Identifies a process other than the Amazon ECS agent performing a DNS lookup for an ECS Agent Communication Service (ACS) or Telemetry Service (TACS) hostname on a Linux host. The ECScape technique abuses the undocumented ACS protocol: a compromised container that can reach the instance metadata service steals the EC2 instance role credentials, then impersonates the ECS agent over ACS to receive the task role credentials of every other task scheduled on the same host. No container escape is required, and the credential theft crosses task boundaries that operators assume are isolated. Only the ECS agent should be speaking this protocol.

Rule type: esql
Rule indices:

Rule Severity: medium
Risk Score: 47
Runs every:
Searches indices from: now-9m
Maximum alerts per execution: 100
References:

Tags:

  • Domain: Endpoint
  • Domain: Cloud
  • Platform: AWS
  • Platform: Linux
  • OS: Linux
  • Tactic: Discovery
  • Data Source: Elastic Defend
  • Rule Type: ES|QL
  • Resources: Investigation Guide

Version: 1
Rule authors:

  • Elastic

Rule license: Elastic License v2

This rule requires data coming in from Elastic Defend.

Elastic Defend is integrated into the Elastic Agent using Fleet. Upon configuration, the integration allows the Elastic Agent to monitor events on your host and send data to the Elastic Security app.

  • Fleet is required for Elastic Defend.
  • To configure Fleet Server refer to the documentation.
  • Go to the Kibana home page and click "Add integrations".
  • In the query bar, search for "Elastic Defend" and select the integration to see more details about it.
  • Click "Add Elastic Defend".
  • Configure the integration name and optionally add a description.
  • Select the type of environment you want to protect, either "Traditional Endpoints" or "Cloud Workloads".
  • Enter a name for the agent policy in "New agent policy name". If other agent policies already exist, you can click the "Existing hosts" tab and select an existing policy instead.
  • Click "Save and Continue".
  • To complete the integration, select "Add Elastic Agent to your hosts" and install Elastic Agent on your hosts. For more details on Elastic Defend refer to the helper guide.

ACS (ecs-a-*.<region>.amazonaws.com) and TACS (ecs-t-*.<region>.amazonaws.com) are the private control and telemetry channels between the ECS agent and the ECS service. No workload process has a legitimate reason to resolve them. A non-agent process doing so is consistent with ECScape, where a compromised task uses stolen instance-role credentials to impersonate the agent over ACS and receive the task role credentials of every other task on the host.

  • Pivot on process.entity_id to the process start event for process.command_line, process.parent.name, and process.parent.executable; network events only carry process.parent.entity_id.
  • Confirm the process is not the ECS agent itself. On ECS-optimized AMIs the agent runs as the ecs-agent container with executable /agent; on Bottlerocket and package installs it is /usr/bin/amazon-ecs-agent.
  • Look for a preceding connection from the same host to 169.254.169.254 or 169.254.170.2 by a non-agent process, which would be the credential theft step.
  • Review CloudTrail for DiscoverPollEndpoint calls made with the instance role from an unexpected source, and for API calls made with other tasks' role credentials shortly after.
  • Check whether the process subsequently spawned children or wrote AWS credentials files.
  • Look for lateral movement indicators: unusual parent processes, shell scripts, or memory-resident execution that would explain why an unexpected binary is contacting the ECS control plane (ACS/TACS).
  • Custom ECS-compatible agents, schedulers, or health checks that speak to ACS/TACS directly may trigger this rule.
  • If the alert fires on a known tool, add its executable path or process name to the rule's exclusion list.
  • Isolate the host if credential theft is suspected and rotate any AWS credentials that may have been exposed.
  • Review CloudTrail for API calls made with the task's IAM role shortly after the event timestamp.
  • Investigate how the suspicious process was introduced and remediate the root cause (e.g., compromised container image, code execution vulnerability).
FROM logs-endpoint.events.network-* METADATA _id, _index, _version
| WHERE host.os.type == "linux"
    AND event.action IN ("lookup_requested", "lookup_result")
    AND process.executable IS NOT NULL
    AND (
      TO_LOWER(dns.question.name) LIKE "ecs-a-*.amazonaws.com*" OR
      TO_LOWER(dns.question.name) LIKE "ecs-t-*.amazonaws.com*"
    )
    AND NOT process.name IN ("amazon-ecs-agent", "ecs-agent", "amazon-ssm-agent", "aws-vpc-cni")
    AND NOT (
      process.executable == "/agent" OR
      process.executable == "/usr/bin/amazon-ecs-agent" OR
      process.executable LIKE "/managed-agents/*" OR
      process.executable LIKE "/usr/libexec/amazon-ecs-*" OR
      process.executable LIKE "/var/lib/ecs/*" OR
      process.executable LIKE "/opt/Elastic/Agent/*"
    )
| KEEP _id, _index, _version, @timestamp, data_stream.namespace, host.id, host.name, user.id, user.name,
    process.entity_id, process.parent.entity_id, process.pid, process.name, process.executable, dns.question.name

Framework: MITRE ATT&CK