Easily map the attack path and prioritize which CVEs are a threat to your organization
Miggo AIGenerate vendor-ready rules for the observed attack patterns, plus reasoning and safe deployment guidance
Miggo AIRoot Cause AnalysisThe vulnerability exists in two API handlers, getBlockAttrs and batchGetBlockAttrs, within kernel/api/attr.go. The provided patch for CVE-2026-72803 clearly shows the addition of access control checks in both of these functions. Specifically, the getBlockAttrs function was modified to include a call to checkBlockPublishAccess, and the batchGetBlockAttrs function was updated to use filterBlockIDsByPublishAccess. Before these changes, both functions would directly fetch block attributes from the database (sql.GetBlockAttrs and sql.BatchGetBlockAttrs respectively) without verifying if the user, especially an anonymous or low-privileged one, was authorized to view them. This allowed for the disclosure of potentially sensitive information from protected documents, as detailed in the vulnerability description. The vulnerable functions are the handlers themselves, as they were the entry points for the unauthorized data access.
batchGetBlockAttrskernel/api/attr.go
getBlockAttrskernel/api/attr.go
| Package Name | Ecosystem | Vulnerable Versions | First Patched Version |
|---|---|---|---|
| github.com/siyuan-note/siyuan/kernel | go | < 0.0.0-20260724093256-229fdffd7e4a | 0.0.0-20260724093256-229fdffd7e4a |
origin/masterCheckAuthCheckAdminRolePrecondition: publish mode enabled (default port 6808); anonymous when Publish.Auth.Enable is false, otherwise any publish reader account. A document is marked publish-forbidden (or password-protected) and contains a block with a memo and a custom attribute set.
Single-block disclosure:
POST http://127.0.0.1:6808/api/attr/getBlockAttrs
{"id":"<BLOCK_ID_IN_PROTECTED_DOC>"}
Returns the block's IAL: name, alias, memo, bookmark, tags, and all custom-* values.
Bulk disclosure:
POST http://127.0.0.1:6808/api/attr/batchGetBlockAttrs
{"ids":["<ID1>","<ID2>","<ID3>"]}
Returns the attribute sets for every supplied ID in one response, with no per-ID authorization.
Control: the sibling getBlockInfo with the same block ID is refused by checkBlockPublishAccess, confirming the boundary is enforced elsewhere and omitted here.
An anonymous reader (publish mode with auth disabled) or any publish RoleReader can read block attributes belonging to publish-forbidden and password-protected documents, including user-authored memos and arbitrary custom-* values. The batch endpoint turns this into a bulk primitive: given a set of block IDs, an attacker retrieves attributes across the whole workspace in a single request. Confidentiality-only.
Call checkBlockPublishAccess in getBlockAttrs before returning, matching getBlockInfo. For batchGetBlockAttrs, apply the check per ID and drop unauthorized entries from the response rather than failing the whole batch.