A heap-based buffer overflow vulnerability exists in the GGUF library's header.n_kv functionality within llama.cpp (commit 18c2e17). This flaw can be triggered by processing a specially crafted .gguf file, potentially allowing an attacker to execute arbitrary code. The vulnerability has a high severity rating with a CVSS score of 8.8.