Loading vulnerability…
CVE-2026-69083 is an SQL injection vulnerability affecting SiYuan versions earlier than 3.7.3. Reader-reachable full-text asset search functionality processes attacker-controlled search-method parameters and regular-expression input using unsafe SQL construction against a read-write SQLite asset-content database. In particular, regular-expression values can break out of intended REGEXP predicates; raw-SQL search functionality is also insufficiently restricted in affected implementations. The flaw can be reached by unauthenticated users when publish authentication is disabled, or by users holding a publish RoleReader token.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
1 valid exploit after Mallory filtered fakes, detection scripts, and README-only repos.
This repository is a small, standalone Python proof-of-concept exploit for CVE-2026-69083 affecting SiYuan versions prior to 3.7.3. The repo contains only two files: a README describing the vulnerability, impact, and usage, and exploit.py implementing the attack with Python standard-library modules only. The exploit targets the web API endpoint POST /api/search/fullTextSearchAssetContent. It abuses unsafe SQL construction in SiYuan's asset-content search logic, where the user-controlled query value is concatenated into a REGEXP clause without escaping. The default exploit path (method 3) sends the payload zzqx') OR 1=1 -- , which closes the quoted REGEXP string, injects a tautology, and comments out the trailing SQL so the backend returns all rows from the asset-content index. The script then parses the JSON response and prints each row's path, name, extension, and content after stripping <mark> tags. A second capability is exposed through the --sql option, which uses method 2 to pass a raw SQL SELECT statement verbatim to the backend. This allows more controlled data extraction from the asset_contents_fts_case_insensitive table, provided the target still permits that code path. The --grep option is a local post-processing filter that narrows displayed results by keyword in content or path. Operationally, this is an unauthenticated web/network exploit rather than a detection script. It does not establish code execution or persistence; its primary purpose is database read/exfiltration from SiYuan's indexed asset content, with the README noting possible broader SQLite abuse such as ATTACH DATABASE depending on backend behavior. The exploit requires a vulnerable SiYuan instance exposed without accessAuthCode, commonly on port 6806.
Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.
Vendor-confirmed product mapping. Mallory continuously reconciles this list against your asset inventory.
5 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.
A critical unauthenticated SQL injection vulnerability in the application's full-text asset-content search API (method 3). Unsafely concatenated regular-expression input can permit an attacker to bypass query filtering and read indexed notebooks, text segments, synchronization logs, and document assets. Because the SQLite handler has read-write access, attackers may also alter index files or delete assets; under particular filesystem-permission conditions, SQLite ATTACH DATABASE functionality could create a path toward code execution on the host.
An unauthenticated or publish-reader-reachable SQL injection vulnerability in SiYuan's fullTextSearchAssetContent endpoint. Method 2 permits arbitrary SQL execution, while method 3 permits SQL injection through an unescaped REGEXP expression. The vulnerable code operates on a read-write, statement-stacking-capable SQLite asset-content database, enabling cross-notebook data disclosure, database modification, and access to ATTACH-reachable files.
A SQL injection vulnerability in SiYuan before version 3.7.3 affecting the fullTextSearchAssetContent endpoint, allowing unauthenticated attackers with publish RoleReader tokens to execute arbitrary SQL against the asset-content database and read, modify, or delete cross-notebook data.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.