EUVD-2026-90423

In camel-ai camel 0.2.91a1, v0.2.91a2 and v0.2.91a3, CodeExecutionToolkit can run model-produced Python code through SubprocessInterpreter without an approval boundary...

CVE-2026-51857

The camel-ai camel framework versions v0.2.91a1 , v0.2.91a2 , and v0.2.91a3 are vulnerable to unauthorized code execution. The CodeExecutionToolkit component can execute Python code generated by a model via the SubprocessInterpreter because it lacks an approval boundary . This allows for the dire...