Description

SiYuan before v3.8.1 fails to filter invisible-tier content from SQL embed blocks, attribute-view keys, and attribute-view backlinks in publish mode. Anonymous readers can enumerate invisible content through these three listing mechanisms despite admin configuration marking content unlisted.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "5.3",
  "pubDate": "2026-08-30T15:16:45.900Z",
  "pubdate": "2026-08-30T15:16:45.900Z",
  "executiveSummary": "This vulnerability is an improper access control flaw within the SiYuan publishing mechanism, specifically affecting versions prior to v3.8.1.\nThe vulnerability allows anonymous, unauthenticated remote attackers to bypass visibility restrictions and enumerate sensitive or 'invisible-tier' content that has been explicitly marked as unlisted by administrators.\nThe issue stems from insufficient server-side filtering within the SQL embed blocks, attribute-view keys, and attribute-view backlinks modules when operating in publish mode.\nImpact includes the unauthorized disclosure of private data, potentially exposing internal notes or sensitive configurations that were intended to remain private.\nThe vulnerability is critical for organizations relying on SiYuan's publishing features to compartmentalize information, as it undermines the confidentiality of the knowledge base architecture.\nNo authentication or specific privileges are required to exploit this flaw, as the attack vector is accessible to any anonymous reader with network access to the public-facing SiYuan instance.",
  "technicalDetails": "The vulnerability resides in the content rendering logic of SiYuan's publish mode, which fails to enforce authorization checks on data retrieved through specific dynamic content components.\nThe root cause is a deficiency in the server-side input/output validation process where SQL embed blocks and attribute-view mechanisms retrieve content from the underlying database without cross-referencing the visibility status of the target objects.\nWhen a user accesses a published page, the application executes queries to populate SQL embed blocks or attribute views. In affected versions, these queries are not scoped to exclude records marked as invisible or unlisted, effectively ignoring the global administrative access control policy.\nThe attack flow follows a predictable pattern: An anonymous attacker identifies the URL structure of a published SiYuan instance. By crafting specific requests that trigger SQL embed blocks or by inspecting the data payloads generated by attribute-view keys and backlinks, the attacker can force the backend to return content that should be restricted.\nSpecifically, when the application renders an attribute-view backlink or key, it fetches associated metadata and note snippets. Because the filtering logic is applied at the client-side presentation layer rather than the data access layer, the sensitive content is serialized and sent to the client browser, where it is easily discoverable through the network traffic or the rendered DOM.\nThis behavior affects versions of SiYuan prior to v3.8.1. The lack of adequate server-side sanitation means that the 'invisible-tier' flag is effectively treated as a UI suggestion rather than a security boundary.\nExploitation requires minimal technical effort, as it leverages the standard functionality of the application to leak information. An attacker does not need to bypass traditional authentication mechanisms, as the vulnerability is inherent to the public-facing content retrieval components.\nPost-exploitation, an attacker can systematically scrape the knowledge base to reconstruct restricted document structures or extract sensitive metadata, which may contain proprietary information, credentials, or sensitive operational notes, severely impacting the confidentiality of the target installation."
}