CVE-2026-23924: Zabbix Agent 2 Information Disclosure

CVE-2026-23924 Overview

CVE-2026-23924 is an Argument Injection vulnerability affecting Zabbix Agent 2's Docker plugin. The vulnerability exists because the plugin does not properly sanitize the docker.container_info parameters when forwarding them to the Docker daemon. An attacker capable of invoking Agent 2 can read arbitrary files from running Docker containers by injecting malicious parameters via the Docker archive API.

Critical Impact
Attackers with access to Zabbix Agent 2 can exploit improper input sanitization to read arbitrary files from Docker containers, potentially exposing sensitive configuration files, credentials, and application data.

Affected Products

Zabbix Agent 2 with Docker plugin enabled
Systems running Docker containers monitored by Zabbix Agent 2
Environments where untrusted users can invoke Zabbix Agent 2 commands

Discovery Timeline

2026-03-24 - CVE-2026-23924 published to NVD
2026-03-25 - Last updated in NVD database

Technical Details for CVE-2026-23924

Vulnerability Analysis

This vulnerability falls under CWE-88 (Improper Neutralization of Argument Delimiters in a Command), commonly known as Argument Injection. The Zabbix Agent 2 Docker plugin processes user-supplied parameters for the docker.container_info functionality without adequate sanitization before passing them to the underlying Docker daemon API.

The attack requires network access and elevated privileges within the Zabbix monitoring infrastructure. When exploited, the vulnerability allows an attacker to bypass intended access controls on the monitored Docker containers, enabling unauthorized file reads through the Docker archive API. This represents a significant confidentiality breach for containerized applications, as attackers can extract sensitive data such as configuration files, environment variables containing secrets, application source code, and database credentials stored within containers.

Root Cause

The root cause of this vulnerability is insufficient input validation in the Zabbix Agent 2 Docker plugin. When the plugin receives docker.container_info parameters, it fails to properly sanitize or validate these inputs before constructing API calls to the Docker daemon. This allows an attacker to inject additional arguments or manipulate the intended API request, specifically targeting the Docker archive API to extract files from containers.

Attack Vector

The attack vector is network-based and requires the attacker to have privileges sufficient to invoke Zabbix Agent 2 commands. The exploitation flow involves:

The attacker identifies a Zabbix Agent 2 instance with the Docker plugin enabled
The attacker crafts a malicious docker.container_info request with injected parameters
The unsanitized parameters are passed to the Docker daemon
The Docker archive API is leveraged to read arbitrary files from running containers
Sensitive data from the target container is exfiltrated to the attacker

The vulnerability does not require user interaction, making it exploitable in automated attack scenarios once initial access to the Zabbix Agent 2 interface is obtained.

Detection Methods for CVE-2026-23924

Indicators of Compromise

Unusual or malformed docker.container_info requests in Zabbix Agent 2 logs
Unexpected Docker archive API calls that do not align with normal monitoring activities
Access patterns to sensitive file paths within containers via the Docker daemon
Anomalous network traffic between Zabbix Agent 2 and Docker daemon containing archive requests

Detection Strategies

Monitor Zabbix Agent 2 logs for suspicious docker.container_info parameter patterns containing path traversal or injection characters
Implement Docker daemon audit logging to detect unauthorized archive API requests
Deploy network monitoring to identify anomalous data transfers between Zabbix Agent 2 and the Docker daemon
Use SentinelOne Singularity to detect file access anomalies within container environments

Monitoring Recommendations

Enable verbose logging on Zabbix Agent 2 to capture all Docker plugin interactions
Configure Docker daemon audit logs to record all archive API calls with full parameter details
Implement alerting for docker.container_info requests containing unexpected characters or path components
Deploy runtime container security monitoring to detect unauthorized file read operations

How to Mitigate CVE-2026-23924

Immediate Actions Required

Review and restrict access to Zabbix Agent 2 to only trusted users and systems
Audit current Zabbix Agent 2 configurations for Docker plugin usage and disable if not required
Implement network segmentation to limit which systems can communicate with Zabbix Agent 2
Apply the vendor patch as soon as it becomes available from Zabbix

Patch Information

Zabbix has tracked this issue in their support system. For the latest patch information and security updates, refer to the Zabbix Support Issue ZBX-27642. Organizations should monitor this resource for official remediation guidance and apply patches according to their change management procedures.

Workarounds

Disable the Docker plugin in Zabbix Agent 2 configuration if Docker container monitoring is not essential
Implement strict network ACLs to limit access to Zabbix Agent 2 to only authorized monitoring infrastructure
Deploy application-layer filtering to inspect and sanitize docker.container_info parameters before they reach the agent
Use read-only container deployments where possible to minimize the impact of unauthorized file reads

bash

# Example: Disable Docker plugin in Zabbix Agent 2 configuration
# Edit /etc/zabbix/zabbix_agent2.conf or zabbix_agent2.d/plugins.d/docker.conf

# Option 1: Disable the Docker plugin entirely
Plugins.Docker.System.Path=

# Option 2: Restrict Docker socket access (if plugin must remain enabled)
# Ensure only the zabbix user can access the Docker socket
chmod 660 /var/run/docker.sock
chown root:zabbix /var/run/docker.sock

# Option 3: Use network policies to restrict Agent 2 access
# iptables example to limit Zabbix Agent 2 connections
iptables -A INPUT -p tcp --dport 10050 -s trusted_server_ip -j ACCEPT
iptables -A INPUT -p tcp --dport 10050 -j DROP