SQLi vulnerability in Cacti could lead to RCE (CVE-2023-51448)

Cacti, a widely used open-source network monitoring and graphing platform, contains a command injection flaw (CWE-74) that lets a completely unauthenticated attacker execute arbitrary commands on the Cacti server. The bug is reachable through the web-facing remote_agent.php script, where an attacker can spoof a trusted poller address (e.g., via a forwarded-client HTTP header) to bypass authentication and inject shell metacharacters into parameters used to build polling commands. Successful exploitation yields code execution in the context of the web server, giving an attacker a foothold on a monitoring system that typically has broad network visibility and stored credentials for many managed devices. Any Cacti deployment running vulnerable code is affected, with internet-exposed instances at greatest risk. Exploitation is confirmed in the wild — the flaw was added to CISA's Known Exploited Vulnerabilities catalog on 2023-02-16, and EPSS assigns it a 99.8% probability of exploitation.

Do: Apply CISA's required action by upgrading Cacti to the patched release (1.2.23 or later per vendor guidance). Until patched, restrict access to remote_agent.php to known poller source addresses and monitor web access and Cacti logs for signs of exploitation. Prioritize internet-exposed Cacti servers, since KEV listing confirms active exploitation and ransomware use is listed as unknown.

Cacti provides an operational monitoring and fault management framework. Version 1.2.25 has a Blind SQL Injection (SQLi) vulnerability within the SNMP Notification Receivers feature in the file `‘managers.php’`. An authenticated attacker with the “Settings/Utilities” permission can send a crafted HTTP GET request to the endpoint `‘/cacti/managers.php’` with an SQLi payload in the `‘selected_graphs_array’` HTTP GET parameter. As of time of publication, no patched versions exist.