Unpatched Critical LMCache Flaw Lets Unauthenticated Attackers Run Code Remotely

A critical vulnerability in LMCache, open-source software that speeds up large language model (LLM) servers such as vLLM, lets an attacker run code on the cache server without logging in, and no fixed version is available. The flaw is in LMCache's multiprocess mode, where the cache runs as a standalone server that LLM workers reach over the ZeroMQ messaging library. A single network

The report states that an unpatched critical vulnerability in the open-source LMCache component can allow unauthenticated remote code execution in multiprocess deployments when the cache server is reachable over a routable network address. The issue affects versions 0.3.9 through 0.5.5, with no fixed release available at the time of reporting. RealGround analysis: because LMCache supports LLM-serving infrastructure, compromise could affect the security of AI workloads and their software supply chain; organizations should inventory affected dependencies, restrict network exposure, and test compensating controls until a patch is released.