CVE-2024-34359 in llama-cpp-python exposes a critical vulnerability through the Jinja2ChatFormatter class. This Server Side Template Injection (SSTI) flaw allows for potential remote code execution due to unsanitized parsing of chat templates within .gguf files. Specifically, the __init__ constructor of the Llama class uses a sandbox-less jinja2.Environment to parse chat templates, which can then be exploited by malicious payloads embedded in the .gguf metadata. Despite a CVSS score of 0, the presence of the "In The Wild" tag highlights active exploitation, but the SOCRadar Vulnerability Risk Score (SVRS) of 30 indicates a lower immediate risk compared to critical vulnerabilities. However, organizations utilizing llama-cpp-python should carefully inspect .gguf files from untrusted sources to mitigate potential exploitation. The successful exploitation of this vulnerability allows attackers to execute arbitrary code on the affected system, posing a significant security risk. Given the potential for remote code execution, developers need to apply proper sanitization techniques to the chat templates to prevent this vulnerability.