SiYuan EUVD-2026-60585

Exploitation requires publish mode to be explicitly enabled on the target SiYuan instance - this is a non-default setting that must be deliberately activated by the workspace owner; default authenticated-only deployments are not affected. The attacker must have network access to the SiYuan API port (default 6806). The CVSS PR:N metric confirms that no credentials are needed once publish mode is active. Exploitation is further bounded by the need to know or enumerate valid block IDs, though no complex prerequisite beyond network reach and an enabled publish-mode configuration is required. Confidentiality impact is limited to block content and structural metadata within the SiYuan workspace; no lateral movement, code execution, or integrity modification is indicated by available data.

The CVSS 4.0 score of 8.7 (AV:N/AC:L/AT:N/PR:N/UI:N/VC:H) accurately reflects the high confidentiality impact, though the AT:N designation slightly overstates universality because publish mode must be explicitly enabled - a non-default configuration that meaningfully narrows the exposed population. No EPSS data was supplied and neither CISA KEV status nor a public proof-of-concept has been identified, so real-world exploitation probability beyond opportunistic scanning is currently unconfirmed. Despite these mitigating signals, the zero-privilege requirement and low attack complexity mean that even a mid-sized municipal IT team in inowroclaw running a publish-mode SiYuan workspace should treat this as urgent, because exposed block content may include sensitive personal or organizational notes whose disclosure could violate confidentiality obligations. There is no integrity or availability impact, making this a pure disclosure risk. Overall, the vulnerability is a genuine priority for any deployment with publish mode active and should be patched promptly; default authenticated-only deployments face no exploitation path from this flaw.

An attacker discovers a network-accessible SiYuan instance running in publish mode and, without any credentials, iterates or guesses block IDs - which follow a known structured format - to query endpoints such as /api/block/getRefText and /api/block/getBlockBreadcrumb via standard HTTP requests. The API returns private block text, cross-reference relationships, and hierarchical breadcrumb paths for blocks the workspace owner intended to keep private. No public proof-of-concept has been identified, but the attack requires only a working HTTP client and no specialist tooling.