
💥 A public exploit is out for CVE-2024-3271, the April 2024 LlamaIndex safe_eval command injection. If your app runs llama-index-core and passes LLM output or user-influenced text into safe_eval, its underscore filter can be bypassed to run OS commands on your server. Apps that never call safe_eval are not affected. First public Apr 16, 2024 · public exploit · fix llama-index-core 0.10.24 · fix llamaindex 0.10.26 https://infrastack.ai/live/CVE-2024-3271?u=1794843479&utm_source=x&utm_medium=social&utm_content=471
