Code Injection in MindsDB OpenBBtable Allows Remote Code Execution
A high-severity flaw in MindsDB lets authenticated attackers run arbitrary Python code via the OpenBBtable class, requiring immediate review of access controls.
Key points
- CVE-2026-92207 carries a CVSS score of 8.8, rated as HIGH severity.
- The vulnerability exists in the OpenBBtable class within MindsDB version 25.10.1.
- Attackers must authenticate to exploit the code injection flaw and execute commands.
An authenticated code injection vulnerability in MindsDB allows remote attackers to execute arbitrary code on affected installations. The flaw resides in the OpenBBtable class, where user-supplied strings are not properly validated before being used to execute Python code. This issue was originally tracked as ZDI-CAN-28388 and affects MindsDB version 25.10.1.
Root cause
The vulnerability stems from inadequate input validation within the OpenBBtable component. When the application processes a string provided by a user, it fails to sanitize or restrict the content before passing it to a Python execution context. This lack of proper validation enables an attacker to inject malicious commands that the service account will then execute, effectively bridging the gap between input and code execution.
Attack path
To exploit this flaw, an adversary must first obtain valid credentials for the MindsDB installation. Once authenticated, the attacker can craft a specific request containing malicious Python code within the input handled by the OpenBBtable class. The application then executes this code without restriction, allowing the attacker to run arbitrary commands in the context of the service account running the MindsDB process.
Affected versions
- MindsDB 25.10.1
Mitigation
- Restrict network access to MindsDB endpoints to trusted internal networks only.
- Enforce strict role-based access controls to limit authenticated users who can interact with the OpenBBtable class.
- Monitor service account permissions to minimize the impact of any successful code execution.
Detection notes
Security operations teams should review logs for unusual Python execution patterns originating from the MindsDB service account. Look for unexpected process spawns or outbound connections that do not align with normal database operations. Check for anomalous input payloads in application logs that contain code-like structures. If suspicious activity is detected, isolate the affected instance immediately to prevent lateral movement while investigating the extent of the compromise.
What to do and how to stay safe: MindsDB
- Audit all user accounts with access to MindsDB and remove unnecessary privileges.
- Monitor system calls and process creation for the MindsDB service account to detect unauthorized code execution.
- Implement network segmentation to limit access to the application from untrusted zones.
- Once the vendor provides an update, apply it promptly to address the injection flaw.
Step-by-step guide: Unpatched Software Myths: Why Common Beliefs Fail to Protect You
General security guidance from the BlueTeam Insider newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
What is the severity of CVE-2026-92207?
The vulnerability has a CVSS score of 8.8, which is rated as HIGH severity.
Is authentication required to exploit this vulnerability?
Yes, an attacker must authenticate to the MindsDB installation before they can exploit the code injection flaw.
Which version of MindsDB is affected by this issue?
MindsDB version 25.10.1 is the specific version identified as affected by this vulnerability.
