Foundation Agents MetaGPT Code Injection Remote Code Execution Vulnerability

A remote code execution vulnerability has been identified in Foundation Agents MetaGPT. This issue arises in the actionoutput_str_to_mapping function, where user-supplied strings are not properly validated before being used to execute Python code. As a result, remote attackers can execute arbitrary code on the affected system, using the service account's privileges. Notably, no authentication is required to exploit this vulnerability.

Impact

Exploitation of this vulnerability allows for arbitrary code execution on the affected system, with the executed code running in the context of the service account.

Remediation

Given the nature of this vulnerability, the recommended mitigation strategy is to limit interaction with the product.