Command Injection Vulnerability in run-llama/llama_index Repository
CVE-2024-3271
9.8CRITICAL
What is CVE-2024-3271?
A command injection vulnerability exists in the Llama_Index repository developed by Run Llama. The flaw is located within the safe_eval function, where attackers exploit the mechanism designed to validate code generated by large language models (LLMs). By providing specially crafted input that lacks underscores, attackers can bypass intended security checks, enabling the execution of arbitrary operating system commands. This vulnerability poses a serious risk of remote code execution on environments where the application is deployed, potentially compromising the integrity and security of the hosting server.
Affected Version(s)
run-llama/llama_index < 10.26