Description

vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.22.1, the vLLM Dockerfile is vulnerable to a dependency confusion attack through the flashinfer-jit-cache package. The package is installed from a custom index (flashinfer.ai/whl/) using --extra-index-url, but the package name was not registered on PyPI, and UV_INDEX_STRATEGY="unsafe-best-match" is set globally. An attacker who registers flashinfer-jit-cache on PyPI with version 0.6.11.post2 can execute arbitrary code as root during the Docker build and backdoor every resulting container image, enabling exfiltration of all user prompts, API credentials, and model data from production vLLM deployments This vulnerability is fixed in 0.22.1.

Severity (CVSS)

Base score8.8
SeverityHigh
VersionCVSS 3.1
VectorCVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
Provided byCNA

Weaknesses

  • CWE-427 — CWE-427: Uncontrolled Search Path Element

Affected products

VendorProductVersions
vllm-projectvllm< 0.22.1

References

Authoritative sources

This page is a snapshot. For the latest enrichment and updates, view the record on CVE.org or the NVD.

Generated from the official CVE List on 23 Jun 2026 10:05 UTC.