SiYuan CVE-2026-72800

Exploitation requires that the SiYuan instance be network-reachable and that the publish feature be active; instances running in fully air-gapped or localhost-only configurations are not reachable by external attackers. The provided CVSS 4.0 vector specifies PR:N, but the CVE description refers to 'authenticated readers' as a prerequisite for getAttributeViewKeysByID access - whether any of the three endpoints truly require no credentials or require publish-reader-level access (which may itself be unauthenticated in SiYuan's model) is not independently confirmed from available data. Impact is additionally bounded by the nature of exposed data: the vulnerability surfaces schema metadata and block identifiers, not raw document bodies; whether block IDs can be directly leveraged to retrieve full content from password-protected documents through additional API calls is not confirmed in available data. Instances where no sensitive content exists behind publish boundaries face negligible practical risk.

The provided CVSS 4.0 vector assigns PR:N (no privileges required), yet the CVE description explicitly references 'authenticated readers' as the threat actor for at least the getAttributeViewKeysByID endpoint - this discrepancy is unresolved in available data and materially affects real-world risk assessment; defenders should confirm the exact authentication requirement per endpoint against the vendor advisory before classifying the exposure. Impact metrics are bounded: VC:L and SC:L indicate only low confidentiality exposure across vulnerable and subsequent systems respectively, with no integrity or availability dimension. Even a mid-sized municipal IT team in sopot running a self-hosted SiYuan instance should treat this as a routine patch cycle rather than an emergency incident, given the narrow confidentiality scope, absence of known active exploitation, and lack of a public exploit. No EPSS data was provided; without that signal, exploitation probability cannot be quantified beyond noting the absence of KEV listing and public POC. The realistic risk concentrates on metadata leakage - schema definitions, vocabulary lists - and block ID enumeration that could facilitate targeted follow-on content discovery, not mass workspace exfiltration.

An attacker with network access to an exposed SiYuan publish interface sends standard HTTP GET requests to getBlockDefIDsByRefText or getBlockRelevantIDs with an arbitrary reference text query, receiving workspace-wide block identifiers that span publish restrictions. Using harvested block IDs, the attacker queries getAttributeViewKeysByID to extract database column schemas - including template expressions and option vocabularies - from attribute view blocks in documents that the publish configuration intended to keep private. No public exploit code is known, making exploitation dependent on manual API interaction or custom tooling.