CVE-2026-51858

15th percentile - higher than 15% of all known CVEs

In camel-ai camel 0.2.91a1, v0.2.91a2 and v0.2.91a3, TerminalToolkit.shell_exec allows prompt-driven shell command execution without an approval boundary. This means the model can initiate execution of arbitrary system commands.

An attacker can prompt the agent to execute arbitrary shell commands, potentially leading to system takeover or data leakage. The lack of approval poses a serious risk in production environments.

Upgrade camel-ai to a version containing an approval mechanism or restrict the ability to invoke TerminalToolkit.shell_exec. Introduce additional controls and isolation for executed commands.