CVE-2024-6825: RCE in litellm Python Library

The vulnerability lies in the way litellm handles the 'postcallrules' configuration, which allows users to define callback functions to be executed after a chat response is processed. An attacker can exploit this by injecting a malicious callback function name. The library splits the provided value at the final '.' mark, treating the last part as the function name and appending '.py'. This allows an attacker to specify a system method, such as 'os.system', as the callback, effectively enabling arbitrary command execution when a chat response is processed. The potential impact is severe, as an attacker could gain complete control over the system running the litellm library, leading to data breaches, system compromise, and further malicious activity.

1. Reserved2024-07-16
2. Published2025-03-20
3. Modified2025-10-16
4. EPSS updated2026-04-02

The primary mitigation for CVE-2024-6825 is to upgrade to version v1.65.4.dev6 or later. If upgrading immediately is not feasible, consider temporarily disabling the 'postcallrules' configuration to prevent the vulnerability from being exploited. Review existing 'postcallrules' configurations for any suspicious or unexpected function names. Implement input validation and sanitization on any user-provided data used in the 'postcallrules' configuration to prevent malicious code injection. Monitor system logs for any unusual activity or command executions related to litellm.