CVE-2025-50857: ZenTaoPMS Path Traversal Vulnerability

CVE-2025-50857 Overview

CVE-2025-50857 is a directory traversal vulnerability affecting ZenTaoPMS versions v18.11 through v21.6.beta. The flaw resides in the /module/ai/control.php component and allows unauthenticated attackers to execute arbitrary code through a crafted file upload. The weakness is classified under CWE-22 (Improper Limitation of a Pathname to a Restricted Directory). Network-based exploitation requires no privileges or user interaction, making the vulnerability suitable for opportunistic attacks against internet-exposed project management deployments.

Critical Impact
Remote attackers can write files to arbitrary locations on the server through /module/ai/control.php, leading to arbitrary code execution and full compromise of ZenTaoPMS instances.

Affected Products

ZenTaoPMS v18.11
ZenTaoPMS versions between v18.11 and v21.6.beta
ZenTaoPMS v21.6.beta

Discovery Timeline

2026-02-26 - CVE-2025-50857 published to NVD
2026-04-15 - Last updated in NVD database

Technical Details for CVE-2025-50857

Vulnerability Analysis

The vulnerability is a directory traversal flaw in the ZenTaoPMS AI module file handler. The /module/ai/control.php endpoint accepts file upload input without correctly validating or canonicalizing the destination path. Attackers can supply traversal sequences such as ../ to escape the intended upload directory and place files in arbitrary locations within the web root.

When the attacker writes a PHP file to an executable directory, the server interprets the file on subsequent requests. This converts a path traversal primitive into arbitrary code execution. Because the endpoint does not require authentication, attackers can reach the vulnerable handler directly over the network.

The EPSS score of 1.166% places this issue in the 78th percentile for exploitation likelihood, reflecting its accessibility to unauthenticated attackers.

Root Cause

The root cause is missing path sanitization in the AI module upload handler. The control script trusts user-supplied filename or path parameters and concatenates them into a destination path without rejecting traversal characters or enforcing an allowlist of upload directories. Standard mitigations such as basename() filtering, canonical path comparison, and extension restrictions are absent on the vulnerable code path.

Attack Vector

An unauthenticated remote attacker sends a crafted HTTP POST request to /module/ai/control.php containing a file upload with a manipulated path. The traversal payload directs the server to write the uploaded file into a web-accessible directory under the ZenTao installation. The attacker then requests the planted file to trigger PHP execution under the privileges of the web server process.

Further technical details and proof-of-concept material are documented in the GitHub Gist Code Snippet and the GitHub Repository Test Code.

Detection Methods for CVE-2025-50857

Indicators of Compromise

Unexpected .php or script files written under ZenTao directories such as www/, tmp/, or module subdirectories with recent modification timestamps.
HTTP POST requests to /module/ai/control.php containing ../ sequences, encoded traversal characters (%2e%2e%2f), or absolute paths in upload parameters.
Web server processes spawning shell interpreters such as sh, bash, cmd.exe, or powershell.exe after requests to the AI module endpoint.

Detection Strategies

Inspect web access logs for requests targeting /module/ai/control.php with multipart upload bodies and review associated user agents and source IPs.
Correlate file creation events under the ZenTao web root with preceding requests to the AI module to identify drop-and-execute patterns.
Hunt for outbound network connections initiated by the PHP-FPM or web server process tree shortly after AI module traffic.

Monitoring Recommendations

Enable file integrity monitoring on the ZenTao installation directory to alert on new executable files.
Forward web server and PHP error logs to a centralized logging platform and retain them for incident review.
Alert on any process execution chain where the web server user invokes interpreters, package managers, or download utilities such as curl and wget.

How to Mitigate CVE-2025-50857

Immediate Actions Required

Restrict network access to ZenTaoPMS instances by placing them behind a VPN or IP allowlist until a fixed release is deployed.
Block or rate-limit requests to /module/ai/control.php at the reverse proxy or web application firewall.
Audit the ZenTao web root for unexpected PHP files and remove any unauthorized artifacts identified during review.

Patch Information

No vendor patch URL is referenced in the published advisory data. Operators should monitor the ZenTaoPMS vendor channels for a release that supersedes v21.6.beta and apply it as soon as it becomes available. Until then, treat all instances running v18.11 through v21.6.beta as vulnerable.

Workarounds

Disable or remove the AI module if it is not required for business operations to eliminate the vulnerable code path.
Deploy a web application firewall rule that rejects upload requests containing ../, ..\\, or URL-encoded traversal sequences targeting the AI control endpoint.
Run the PHP worker under a least-privilege account with write access limited to directories that are not served by the web server.

bash

# Example WAF/nginx rule to block traversal payloads to the AI module
location /module/ai/control.php {
    if ($request_uri ~* "(\.\./|\.\.\\|%2e%2e%2f|%2e%2e/)") {
        return 403;
    }
    client_max_body_size 2m;
}