A newly disclosed chain of pre-authentication remote code execution vulnerabilities in CyberPanel could enable unauthenticated attackers to compromise exposed servers.
The attack, named ShadowPanel by Pentera researcher Nir Chako, combines weaknesses in the platform’s AI Scanner component with its built-in cron-job functionality to move from unauthenticated access to host-level command execution.
CyberPanel is a web hosting control panel for managing websites, DNS configurations, email accounts, databases, users, and scheduled tasks.
CyberPanel Flaws Chain Authentication Bypass
A successful compromise is therefore especially significant because the software operates as an administrative layer over services hosted on a server.
An attacker with control-panel privileges can potentially alter websites, create user accounts, access databases, and establish persistent access.
The ShadowPanel chain affects CyberPanel version 2.4 and relies on multiple weaknesses that are more dangerous together than individually.
The initial stage targets exposed AI Scanner interfaces. Researchers found that unauthenticated debug API routes, including /api/ai-scanner/list-api-keys, could disclose information about recent scan activity.
This information included valid scan identifiers, known as scan_id values. These identifiers are central to the attack because CyberPanel’s AI Scanner callback mechanism requires a valid scan ID before accepting scan results.
Once an attacker obtains a legitimate identifier, they can interact with vulnerable worker endpoints and submit attacker-controlled data to the application.
CVE-2026-41473 tracks the authentication bypass in CyberPanel’s AI Scanner worker API. The flaw affects endpoints such as /api/ai-scanner/status-webhook and /api/ai-scanner/callback, which did not adequately verify that incoming requests were authorized.
As a result, a remote attacker could write arbitrary content into scan-related database entries without possessing a CyberPanel account or valid administrator credentials.
The next stage involves CVE-2026-41472, a stored cross-site scripting vulnerability in the AI Scanner dashboard. Attackers can use the unauthenticated callback capability to inject malicious JavaScript into forged scan findings.
CyberPanel stores those findings in its database and later displays them in the AI Scanner interface without properly escaping untrusted content.
When an administrator opens the manipulated scan results, the malicious script executes inside the administrator’s authenticated browser session. This allows the attacker to perform actions via the victim’s active CyberPanel session.
Pentera’s proof of concept used this access to create a new CyberPanel administrator account, turning a browser-based XSS flaw into persistent privileged access.
The final stage abuses CyberPanel’s legitimate cron-job management capabilities. Cron jobs are designed to execute scheduled operating-system commands, making them a powerful feature for administrators and a high-impact target for attackers.
After obtaining privileged panel access, the researchers created scheduled tasks containing commands that resulted in a reverse shell on the vulnerable server.
The chain demonstrates why the severity of vulnerability cannot always be evaluated in isolation. Exposed debug interfaces, unauthenticated database writes, stored XSS, and privileged administrative functions can collectively form a complete attack path.
Pentera reported the issue to CyberPanel on May 4, 2026. CyberPanel reportedly issued fixes within approximately four hours by removing debug endpoints, adding callback authentication, and implementing output escaping.
Administrators should immediately upgrade to the latest patched CyberPanel release. Until patching is complete, organizations should disable AI Scanner where possible.
Restrict access to the management panel, block vulnerable AI Scanner API paths, and investigate any unexpected callbacks, altered scan results, newly created administrator accounts, and unauthorized cron jobs.
Give your security team the visibility and context to investigate suspicious activity faster and contain threats before business impact grows. Strengthen Your Investigations with ANY.RUN
