SiYuan CVE-2026-72799

Exploitation requires SiYuan to be running with publish mode active and the Publish.Auth.Enable configuration flag set to false. Under this condition, any anonymous (unauthenticated) HTTP client, or any client presenting a publish reader token of any privilege level, can call the five affected endpoints. Exploitation is blocked when Publish.Auth.Enable is true (auth enabled), when the SiYuan HTTP port is not reachable from the attacker's network, or when the instance is running without publish mode activated. The vulnerability does not affect SiYuan instances used in a purely local or private-network context with no publish exposure. Authentication requirements are confirmed by the CVSS 4.0 PR:N metric in the provided vector.

The CVSS 4.0 score of 6.9 (AV:N/AC:L/AT:N/PR:N/UI:N) reflects that the vulnerability is remotely exploitable with no prerequisites on a network-accessible publish server. Impact is bounded to partial confidentiality - document metadata and structure are exposed, but body content of protected documents is not directly returned by these endpoints. No EPSS data was supplied; given no KEV listing and no reported public exploit, widespread opportunistic scanning is not confirmed. The practical risk is context-dependent: for users who intentionally expose SiYuan's publish interface to the internet with auth disabled, the attack surface is real and requires no exploitation skill. For users running in a private network with auth enabled, there is no exposure. The CVSS score is appropriate for the information-disclosure class of impact, but organizational sensitivity of the document tree (e.g., internal project structure, confidential notebook names) can elevate business risk well beyond the numeric score.

An attacker discovers a SiYuan instance exposed to the internet in publish mode with authentication disabled. Without any credentials, the attacker sends HTTP requests to the getFullHPathByID and getHPathByID endpoints, iterating or guessing document IDs to reconstruct the full notebook hierarchy - including the names and paths of documents the owner marked as hidden or password-protected. The harvested document IDs and path mappings can then be used to target subsequent access attempts against other SiYuan API endpoints that may serve content. No public exploit code has been identified, but exploitation requires only standard HTTP tooling.