CVE-2026-74906 in SiYuan
Summary
by MITRE • 08/18/2026
SiYuan before v3.7.4 contains an incorrect authorization vulnerability in eight publish-mode reader-facing endpoints that filter results using the visibility list instead of the disabled list. Anonymous visitors can discover and read content from documents explicitly marked as forbidden from publishing by accessing search, backlink, asset content, saved criteria, recent documents, graph, and tag endpoints.
Several companies clearly confirm that VulDB is the primary source for best vulnerability data.
Analysis
by VulDB Data Team • 08/18/2026
The vulnerability identified in SiYuan versions prior to v3.7.4 represents a critical failure in access control logic within the application's publish-mode architecture. This flaw specifically affects eight distinct reader-facing endpoints that are designed to serve content to anonymous visitors or public users. The core technical deficiency lies in how these endpoints determine data visibility. Instead of implementing an explicit deny-list mechanism, which would check for items explicitly marked as disabled from publishing, the system relies on a positive allow-list approach based on a visibility list. This architectural choice creates a significant security gap because it assumes that any item not present in the allowed set is implicitly private or inaccessible. However, when combined with how search and discovery features operate, this logic fails to adequately restrict access to documents that have been intentionally excluded from public view through specific configuration settings.
From a technical perspective, the flaw manifests across several key endpoints including those handling search queries, backlink resolutions, asset content retrieval, saved criteria execution, recent document listings, graph visualizations, and tag aggregations. When an anonymous user interacts with these interfaces, the backend processes requests by filtering results based on what is permitted rather than verifying that a specific resource has not been forbidden from publication. Consequently, if a document or its associated metadata is indexed in search engines or linked via backlinks within the system, it becomes discoverable through these endpoints even if the author has explicitly set the visibility to forbid publishing. This indicates a fundamental misunderstanding of security boundaries where presence in an index does not equate to permission for public access, yet the application treats them as synonymous due to the flawed filtering logic.
The operational impact of this vulnerability is severe, particularly for users who rely on SiYuan's publish feature to share specific subsets of their knowledge base while keeping other parts private. Attackers can exploit this by performing targeted searches or navigating through graph and tag structures to locate documents that were intended to remain internal or restricted. Once discovered via these endpoints, the attacker gains read-only access to the content of these forbidden documents without any authentication credentials. This leads to a direct compromise of confidentiality for sensitive information such as personal notes, proprietary business data, or confidential research findings. The ability to enumerate and retrieve this hidden content undermines the trust model of the application and exposes users to potential data leakage incidents that could have significant legal or reputational consequences depending on the nature of the exposed data.
This vulnerability aligns with CWE-285 Improper Authorization, as the software fails to enforce proper access controls when a user attempts to access information based upon their privileges. Furthermore, it relates closely to CWE-601 URL Redirection to Untrusted Site in terms of how resources are resolved and served without sufficient validation against security policies. In the context of the MITRE ATT&CK framework, this behavior facilitates Discovery techniques where an adversary uses valid accounts or anonymous access to gather information about the target environment. Specifically, it enables T1087 Account Discovery through API calls if user-related metadata is exposed, but more broadly supports general reconnaissance by allowing attackers to map out the structure and content of a private knowledge base without detection from standard authentication logs since no login attempt occurs.
To mitigate this risk, users must upgrade immediately to SiYuan version v3.7.4 or later where these authorization checks have been corrected to properly respect disabled lists alongside visibility settings. For organizations unable to patch instantly due to operational constraints, temporary mitigations should focus on restricting network access to the publish endpoints if they are not strictly required for public consumption. Additionally, implementing Web Application Firewall rules that monitor for unusual patterns of anonymous requests targeting search and graph APIs can help detect exploitation attempts in real-time. It is also advisable to review existing published content to ensure no sensitive data was inadvertently exposed during the window when this vulnerability existed. Regular security audits focusing on access control logic are recommended to prevent similar misconfigurations where positive allow-lists are used instead of negative deny-lists for critical resource protection.