SQL Injection Flaw Impacts SiYuan App

A severe security vulnerability has been identified within the SiYuan note-taking platform, specifically affecting versions prior to v3.7.3. This flaw exposes the backend database to unauthorized interaction, potentially allowing an attacker to read or modify cleartext content stored within the application's notebooks.

Understanding the Search Endpoint Flaw

The vulnerability exists within the /api/filetree/searchDocs endpoint, which handles incoming requests to search through stored documents. Analysis shows that the keyword parameter provided by the caller is directly concatenated into SQL statements without any form of input sanitization, such as escaping or parameter binding.

Because the application utilizes a database driver that supports stacked statements—which are delimited by semicolons—the lack of input filtering allows for the execution of unauthorized commands. An attacker can leverage this to manipulate the database on a read-write SQLite handle, impacting all non-encrypted data contained within the instance.

Access and Exploitation Requirements

The accessibility of this flaw depends on the configuration of the target SiYuan instance. The endpoint can be reached by a user holding a publish RoleReader token. More significantly, the system is exposed to unauthenticated users if the instance has been configured with publish mode enabled and the Publish.Auth.Enable setting explicitly toggled to false.

The impact is categorized as Critical, reflecting the ease of remote exploitation and the high level of access provided to an attacker. The vulnerability does not require user interaction to be triggered, and the network-based attack vector allows for exploitation without pre-existing privileges in many deployment scenarios.

Quantitative Risk Assessment

• CVE Identifier: CVE-2026-69085
• CVSS 3.1 Base Score: 10
• Affected Versions: Before v3.7.3
• Publication Date: 2026-08-03

Implications for Data Integrity

This vulnerability highlights the risks associated with improper handling of user-supplied data in backend API calls. For users running self-hosted versions of SiYuan, the ability of an external actor to perform unauthorized read and write operations on notebook data represents a significant threat to information privacy and integrity.

Organizations or individuals relying on SiYuan should ensure their instances are updated to at least version v3.7.3 to mitigate the risk. Maintaining strict authentication policies and limiting exposure of API endpoints to untrusted networks remains a critical component of securing such deployments against potential exploitation of this nature.