Description

SiYuan versions before v3.8.2 contain an information disclosure vulnerability in the getAttributeViewKeysByID endpoint that allows publish readers to enumerate private attribute view key definitions without verifying parent database visibility. Attackers can access the endpoint to retrieve complete key schemas including sensitive field names and relation definitions from hidden databases.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "4.3",
  "pubDate": "2026-09-05T13:18:14.300Z",
  "pubdate": "2026-09-05T13:18:14.300Z",
  "executiveSummary": "SiYuan versions prior to v3.8.2 are susceptible to an information disclosure vulnerability residing within the getAttributeViewKeysByID endpoint.\nThis flaw stems from inadequate authorization checks, specifically failing to validate the visibility of a parent database before exposing its schema metadata.\nUnauthenticated or low-privileged 'publish readers' can exploit this mechanism to bypass access controls and enumerate private attribute view key definitions.\nThe vulnerability poses a significant risk to data confidentiality, as it facilitates the unauthorized extraction of internal database structures, including sensitive field naming conventions and intricate relation definitions associated with hidden or restricted databases.\nNo complex exploitation requirements are noted, as the endpoint can be queried directly to return schema information that should remain isolated from external viewers.",
  "technicalDetails": "The vulnerability exists within the application's API layer, specifically at the getAttributeViewKeysByID function. The root cause is a breakdown in the object-level authorization logic; the system fails to perform a recursive or hierarchical permission check when processing requests for attribute view keys.\nIn the affected versions, the endpoint relies solely on the ID provided in the request without verifying if the requesting user (specifically a 'publish reader' role) has read permissions for the parent database container associated with those keys.\nThe attack flow proceeds as follows: An attacker identifies the target endpoint, getAttributeViewKeysByID, and crafts a request using the identifier of a private or hidden database attribute view. Because the application logic lacks an explicit check to confirm the user's authorization level relative to the parent object, the server processes the request and returns the requested metadata in a JSON response.\nThe payload returned by the server includes comprehensive schema details. This metadata typically encompasses internal field labels, data types, and complex relation definitions that map the architectural connections between disparate database tables. Even if the actual row data is not exposed, the disclosure of this structural information is highly damaging.\nExposure of these definitions allows an attacker to map the internal data model of the SiYuan instance. This reconnaissance is critical for further exploitation, as it reveals the nature and sensitivity of stored information, enabling targeted data exfiltration if additional vulnerabilities are discovered or if the attacker can guess valid records based on the exposed schema.\nThe vulnerability is accessible over the network via standard HTTP/HTTPS protocols and requires no administrative privileges, as the 'publish reader' role is sufficient to trigger the flaw. The lack of validation creates a direct information leakage vector, as the system treats the request as a legitimate query for metadata despite the sensitive nature of the underlying data source."
}