### [CVE-2024-34359](https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2024-34359)
![](https://img.shields.io/static/v1?label=Product&message=llama-cpp-python&color=blue)
![](https://img.shields.io/static/v1?label=Version&message=%3D%20%3E%3D%200.2.30%2C%20%3C%3D%200.2.71%20&color=brighgreen)
![](https://img.shields.io/static/v1?label=Vulnerability&message=CWE-76%3A%20Improper%20Neutralization%20of%20Equivalent%20Special%20Elements&color=brighgreen)

### Description

llama-cpp-python is the Python bindings for llama.cpp. `llama-cpp-python` depends on class `Llama` in `llama.py` to load `.gguf` llama.cpp or Latency Machine Learning Models. The `__init__` constructor built in the `Llama` takes several parameters to configure the loading and running of the model. Other than `NUMA, LoRa settings`, `loading tokenizers,` and `hardware settings`, `__init__` also loads the `chat template` from targeted `.gguf` 's Metadata and furtherly parses it to `llama_chat_format.Jinja2ChatFormatter.to_chat_handler()` to construct the `self.chat_handler` for this model. Nevertheless, `Jinja2ChatFormatter` parse the `chat template` within the Metadate with sandbox-less `jinja2.Environment`, which is furthermore rendered in `__call__` to construct the `prompt` of interaction. This allows `jinja2` Server Side Template Injection which leads to remote code execution by a carefully constructed payload.

### POC

#### Reference
- https://github.com/abetlen/llama-cpp-python/security/advisories/GHSA-56xg-wfcc-g829

#### Github
No PoCs found on GitHub currently.