CVE-2026-30860
Published: 07 March 2026
Description
WeKnora is an LLM-powered framework designed for deep document understanding and semantic retrieval. Prior to version 0.2.12, a remote code execution (RCE) vulnerability exists in the application's database query functionality. The validation system fails to recursively inspect child nodes within…
more
PostgreSQL array expressions and row expressions, allowing attackers to bypass SQL injection protections. By smuggling dangerous PostgreSQL functions inside these expressions and chaining them with large object operations and library loading capabilities, an unauthenticated attacker can achieve arbitrary code execution on the database server with database user privileges. This issue has been patched in version 0.2.12.
Mitigating Controls (NIST 800-53 r5)AI
SI-10 directly addresses the core issue by requiring validation of database query inputs, including recursive inspection of PostgreSQL array and row expressions to block SQL injection bypasses.
SI-2 mandates timely flaw remediation, such as upgrading WeKnora to version 0.2.12, which patches the validation failure enabling RCE.
AC-6 limits the impact of RCE by enforcing least privilege on the database user, restricting abilities like library loading and large object operations required for full exploitation.
Security SummaryAI
CVE-2026-30860 is a remote code execution (RCE) vulnerability in WeKnora, an LLM-powered framework for deep document understanding and semantic retrieval. The issue affects versions prior to 0.2.12 and resides in the application's database query functionality, which relies on PostgreSQL. The validation system does not recursively inspect child nodes within PostgreSQL array expressions and row expressions, enabling attackers to bypass SQL injection protections (CWE-89). This allows smuggling of dangerous PostgreSQL functions, chained with large object operations and library loading capabilities, resulting in arbitrary code execution on the database server under database user privileges. The vulnerability carries a CVSS v3.1 base score of 9.9 (AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H).
An attacker with low privileges required (per CVSS) can exploit this vulnerability remotely over the network with low complexity and no user interaction. Although described as exploitable by unauthenticated attackers, the scenario involves crafting malicious inputs to the database query endpoint that evade validation, smuggling payloads within array or row expressions. Successful exploitation grants RCE on the PostgreSQL database server, executing arbitrary code with the privileges of the database user running the WeKnora application.
The GitHub Security Advisory (GHSA-8w32-6mrw-q5wv) at https://github.com/Tencent/WeKnora/security/advisories/GHSA-8w32-6mrw-q5wv confirms the issue and states it has been patched in WeKnora version 0.2.12. Security practitioners should upgrade to version 0.2.12 or later to mitigate the vulnerability.
WeKnora's use of large language models for semantic retrieval highlights its relevance to AI/ML deployments, where document processing pipelines may inadvertently expose high-severity database flaws. No real-world exploitation has been reported in available data.
Details
- CWE(s)
Affected Products
AI Security AnalysisAI
- AI Category
- Other AI Platforms
- Risk Domain
- N/A
- OWASP Top 10 for LLMs 2025
- None mapped
- MITRE ATLAS Techniques
- None mapped
- Classification Reason
- Matched keywords: llm
MITRE ATT&CK Enterprise TechniquesAI
Why these techniques?
The vulnerability is a remote SQL injection leading to arbitrary code execution in a network-accessible application (WeKnora with PostgreSQL backend), directly enabling exploitation of public-facing applications.