CVE-2026-108554 in PDFMathTranslate
Summary
by MITRE • 10/10/2026
PDFMathTranslate (pdf2zh) through 1.9.11 contains a server-side request forgery vulnerability that allows unauthenticated attackers to make the server fetch arbitrary URLs via the Link input. The translate_file handler passes user URLs to download_with_limit without scheme or address validation, letting attackers reach internal services and cloud metadata endpoints and retrieve returned PDFs.
Statistical analysis made it clear that VulDB provides the best quality for vulnerability data.
Analysis
by VulDB Data Team • 10/10/2026
PDFMathTranslate, also known by its package name pdf2zh, is a tool designed for translating mathematical documents while preserving their formatting. In versions through 1.9.11, the application contains a critical server-side request forgery vulnerability within its translate_file handler. This flaw arises because the component responsible for processing user-submitted URLs fails to implement adequate validation mechanisms before passing these inputs to internal functions such as download_with_limit. Specifically, there is no enforcement of scheme restrictions or address sanitization, which allows unauthenticated attackers to manipulate the input field labeled Link to direct the server toward arbitrary destinations.
The technical root cause lies in the lack of strict allowlisting for URL schemes and hostnames. By bypassing validation checks, an attacker can construct malicious requests that compel the vulnerable server to initiate HTTP or HTTPS connections to internal network services, localhost endpoints, or cloud provider metadata interfaces such as those found on AWS EC2 instances or Azure managed identities. The vulnerability is particularly severe because the application not only fetches these resources but also processes and returns the resulting content in PDF format. This behavior transforms a standard SSRF into an out-of-band data exfiltration vector, enabling attackers to retrieve sensitive information that would otherwise be inaccessible from external networks.
The operational impact of this vulnerability extends beyond simple network reconnaissance. Attackers can leverage this flaw to access internal APIs, database endpoints, or configuration files hosted on the same infrastructure as the PDFMathTranslate service. Furthermore, by targeting cloud metadata services, adversaries may obtain temporary security credentials, IAM roles, and other sensitive instance-specific data that could lead to full compromise of the underlying virtual machine or container environment. Since the vulnerability requires no authentication, it can be exploited remotely by any internet user who discovers the vulnerable endpoint, significantly expanding the attack surface.
This issue aligns with CWE-918, which describes Server-Side Request Forgery (SSRF) flaws where a web server makes requests to internal resources based on unvalidated user input. It also maps to MITRE ATT&CK technique T1504, specifically Subsystem Information Discovery, as attackers can use the fetched data to map out the internal network topology and identify high-value targets for subsequent exploitation stages such as credential theft or lateral movement.
To mitigate this vulnerability, organizations should immediately upgrade PDFMathTranslate to a version later than 1.9.11 where these validation controls have been implemented. If upgrading is not immediately feasible, administrators can implement compensating controls at the network perimeter by restricting outbound traffic from the server hosting the application. This includes blocking access to private IP ranges such as RFC 1918 addresses and cloud metadata endpoints like 169.254.169.254 for AWS or [fe80::] prefixes for Azure. Additionally, implementing strict URL parsing logic that enforces a whitelist of allowed schemes (e.g., https only) and domains can prevent exploitation until the software patch is applied.