Approved changes feed: RSS · Atom

cpe:2.3:a:abetlen:llama-cpp-python:*:*:*:*:*:*:*:*

part: a version: * update: *

VendorAbetlen (9c3adea2-2d0d-5278-81b4-562a3e1b96e9)
ProductLlama Cpp Python (6596b0e7-7251-5358-b7b3-811de477c477)
Edition*
Language*
Software edition*
Target software*
Target hardware*
Other*
NotesImported from gcve-enriched-dumps CVE data

PURL mappings

PURLSourceLast updated
No PURL mappings for this CPE yet.

Vulnerability references

IdentifiercpeApplicabilitySubmitteddb.gcve.eu detailsRationale
CVE:CVE-2024-34359 vulnerable 2026-06-03 14:55:53.924661 llama-cpp-python vulnerable to Remote Code Execution by Server-Side Template Injection in Model Metadata
CRITICAL (9.7)
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.
Published: 2024-05-10T17:07:18.850Z
Updated: 2024-08-02T02:51:10.739Z
Reference links
Imported from gcve-enriched-dumps CVE data

Contribute

You can submit an edit proposal for this CPE entry or suggest a related product/vendor addition using the action button above.