Description

SiYuan versions before v3.7.4 fail to apply publish-access filtering to the getAttributeViewKeysByID endpoint, allowing authenticated readers to retrieve complete database column schemas including descriptions, select vocabularies, and template expressions. Additionally, getBlockDefIDsByRefText and getBlockRelevantIDs endpoints enumerate workspace-wide block IDs without publish scoping, enabling attackers to discover valid block identifiers across publish boundaries and access content from hidden or password-protected documents.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "5.8",
  "pubDate": "2026-08-12T20:17:51.830Z",
  "pubdate": "2026-08-12T20:17:51.830Z",
  "executiveSummary": "An authorization bypass vulnerability exists in SiYuan versions prior to v3.7.4, specifically involving inadequate publish-access filtering across multiple API endpoints.\nThe vulnerability allows authenticated readers to bypass intended publish boundaries, exposing sensitive workspace-wide metadata, database column schemas, select vocabularies, template expressions, and valid block identifiers.\nAttackers with standard reader privileges can leverage this flaw to enumerate internal database schemas and discover valid block identifiers belonging to hidden or password-protected documents.\nThe impact includes unauthorized information disclosure and potential reconnaissance capabilities that undermine privacy controls and access boundaries enforced by the application.\nExploitation requires low-privileged authenticated access to the target SiYuan instance, enabling users with baseline viewing privileges to harvest unauthorized structural data and internal identifiers across the workspace.",
  "technicalDetails": "The root cause of the vulnerability stems from insufficient access control enforcement and missing publish-access filtering within specific backend API endpoints in SiYuan versions prior to v3.7.4.\nSpecifically, the getAttributeViewKeysByID endpoint fails to apply publish-access restrictions, allowing authenticated readers to query and retrieve complete database column schemas, which include internal descriptions, select vocabularies, and template expressions.\nAdditionally, the getBlockDefIDsByRefText and getBlockRelevantIDs endpoints lack proper publish scoping, enabling users to enumerate workspace-wide block IDs indiscriminately.\nThe attack flow proceeds as follows: an authenticated attacker with standard reader privileges directly interacts with the vulnerable endpoints without triggering proper authorization checks for published versus unpublished or protected content.\nBy invoking the getAttributeViewKeysByID endpoint, the attacker extracts detailed database structures that should normally be restricted.\nConcurrently, by querying getBlockDefIDsByRefText and getBlockRelevantIDs, the attacker harvests valid block identifiers across publish boundaries.\nArmed with these discovered block identifiers, the attacker can subsequently target hidden or password-protected documents to access restricted content.\nThe vulnerable components comprise the backend routing and controller logic handling attribute view retrieval and block reference identification.\nThe affected versions are all SiYuan releases prior to v3.7.4.\nThe prerequisites for this attack include network access to the SiYuan application and a valid low-privileged reader authentication session."
}