This vulnerability exposes a systematic authorization gap in SiYuan's block attribute retrieval endpoints. The core issue is that getBlockAttrs and batchGetBlockAttrs were treated as metadata plumbing rather than data access points, inheriting trust from the document rendering pipeline instead of enforcing their own publish-access checks. The fix in v3.7.4 corrected this for these specific endpoints, but the pattern is the real concern — these auxiliary endpoints exist to reduce client-side complexity (batch operations, preview generation, metadata retrieval) and almost always assume authorization is inherited rather than explicitly enforced. The data at risk is particularly sensitive: block names, aliases, memos, and custom fields encode organizational context — 'this block relates to Project Phoenix,' 'this person is a key stakeholder,' 'internal only.' An attacker who enumerates these attributes across many documents reconstructs not just what information exists, but how the organization categorizes and values that information. That's a knowledge graph, not a data leak. The CVSS 5.8 captures the technical access pattern but fundamentally misrepresents the intelligence value. A medium-rated CVE that reveals organizational knowledge hierarchy is more valuable to an attacker than a high-rated CVE exposing a single document body. The version history (fixed in v3.7.4) tells you when the gap was closed, not when it was introduced — it may have existed from initial feature scope, been stripped during refactoring, or never merged from a parallel branch. Regardless of origin, treat this as evidence that other attribute-adjacent endpoints in SiYuan's API may lack publish-access enforcement. Audit your block-level metadata endpoints for similar gaps, not just the ones explicitly patched. Organizations using SiYuan for sensitive knowledge management should treat this as a signal that their publish-access model may have structural blind spots at the metadata layer.