CVE-2026-69083: SiYuan SQL Injection Vulnerability

CVE-2026-69083 Overview

CVE-2026-69083 is a SQL injection vulnerability in SiYuan, an open-source personal knowledge management application. Versions before v3.7.3 expose the fullTextSearchAssetContent endpoint to unauthenticated users and publish RoleReader tokens. Attackers can inject arbitrary SQL through unescaped method parameters and REGEXP clauses. The injection targets the read-write asset-content database, enabling cross-notebook data access. The flaw is classified under CWE-89 (Improper Neutralization of Special Elements used in an SQL Command).

Critical Impact
Unauthenticated attackers can execute arbitrary SQL to read, modify, or delete data across all notebooks in a SiYuan instance.

Affected Products

SiYuan versions before v3.7.3
SiYuan instances exposing the fullTextSearchAssetContent endpoint
SiYuan deployments publishing RoleReader tokens

Discovery Timeline

2026-08-03 - CVE-2026-69083 published to NVD
2026-08-03 - Last updated in NVD database

Technical Details for CVE-2026-69083

Vulnerability Analysis

The vulnerability resides in SiYuan's fullTextSearchAssetContent endpoint. This endpoint accepts user-supplied parameters and forwards them into SQL statements without proper escaping. The method parameter and REGEXP clauses are directly concatenated into queries executed against the asset-content database.

Because the endpoint is reachable without authentication, any network-adjacent attacker can send crafted requests. SiYuan additionally publishes RoleReader tokens, further reducing the barrier to exploitation. The asset-content database backing the endpoint is opened in read-write mode. Attackers can therefore issue INSERT, UPDATE, and DELETE statements rather than only reading records.

Root Cause

The root cause is improper neutralization of special SQL elements in request parameters. Rather than using parameterized queries or prepared statements, the endpoint constructs SQL through string concatenation. The REGEXP clause pathway compounds the issue by accepting attacker-controlled expressions that can be broken out of the intended pattern context.

Attack Vector

Exploitation requires only network access to the SiYuan instance. An attacker sends an HTTP request to the fullTextSearchAssetContent endpoint with a malicious method value or crafted REGEXP payload. The injected SQL executes with the database's full privileges on the asset-content store. Attackers can enumerate notebook contents, tamper with stored assets, or destroy data across notebook boundaries.

See the GitHub Security Advisory GHSA-fph3-ghq9-vw66 and the VulnCheck SQL Injection Advisory for additional technical detail.

Detection Methods for CVE-2026-69083

Indicators of Compromise

Unexpected HTTP requests to the fullTextSearchAssetContent endpoint from unknown sources
Request parameters containing SQL keywords such as UNION, SELECT, INSERT, UPDATE, or DROP within method or REGEXP fields
Unexplained modifications or deletions of asset content across multiple notebooks
Presence of RoleReader tokens in access logs preceding anomalous queries

Detection Strategies

Inspect application and reverse-proxy logs for requests to fullTextSearchAssetContent containing SQL metacharacters such as single quotes, semicolons, or comment sequences
Alert on unauthenticated calls to SiYuan API endpoints when access should be restricted
Monitor asset-content database files for unexpected write activity or size changes

Monitoring Recommendations

Enable verbose HTTP request logging on SiYuan and any upstream proxy
Forward SiYuan logs to a centralized log platform for correlation and retention
Track outbound connections from the SiYuan host that may indicate data exfiltration following injection

How to Mitigate CVE-2026-69083

Immediate Actions Required

Upgrade SiYuan to version 3.7.3 or later immediately
Restrict network exposure of the SiYuan service to trusted networks or place it behind an authenticating reverse proxy
Rotate any RoleReader tokens that may have been published or exposed
Review asset-content databases for signs of unauthorized modification or deletion

Patch Information

The SiYuan maintainers addressed the flaw in version 3.7.3. The fix is documented in the GitHub Security Advisory GHSA-fph3-ghq9-vw66. Administrators should upgrade all instances to v3.7.3 or later.

Workarounds

Block external access to the fullTextSearchAssetContent endpoint using web application firewall (WAF) rules until patching is complete
Require authentication at a reverse proxy layer for all SiYuan API endpoints
Disable public publishing of notebooks that rely on RoleReader tokens until the instance is patched

bash

# Example nginx rule to block the vulnerable endpoint until patched
location ~* /api/search/fullTextSearchAssetContent {
    return 403;
}