PYSEC-2026-227 CVE-2026-54232 GHSA-jrf6-vqxq-pjv2
vLLM is an inference and serving engine for large language models (LLMs).
- Severity
- 8.8 / 10
- High · CVSS 3.1
- Exploit chance (EPSS)
- 0.56%
- Chance of exploitation in the next 30 days. Higher than 46.81% of scored vulnerabilities.
- Known exploited
- No
- Not in CISA's catalog
- Weakness
- CWE-427 Uncontrolled Search Path Element
- Fix
0.22.1- Upgrade vllm to this version or later
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.
Affected versions
PyPI
| Package | Vulnerable | Fixed in |
|---|---|---|
| vllm | < 0.22.1 |
0.22.1 |
How to fix
Move to the fixed version or later with your package manager, update the lockfile, and run your tests. If the package only arrives through another dependency, upgrade that dependency first. If none of its releases pulls a fixed version yet, force the version with your package manager's override.
vllm PyPI → 0.22.1
- pip
pip install --upgrade "vllm>=0.22.1"- poetry
poetry add "vllm>=0.22.1"- uv
uv add "vllm>=0.22.1"
Severity breakdown
CVSS 3.1 8.8
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
- Attack vector
- Network
- Attack complexity
- Low
- Privileges required
- None
- User interaction
- Required
- Scope
- Unchanged
- Confidentiality
- High
- Integrity
- High
- Availability
- High
Weakness
CWE-427: Uncontrolled Search Path Element
The product uses a fixed or controlled search path to find resources, but one or more locations in that path can be under the control of unintended actors.
How to prevent it
- Architecture and Design, Implementation: Hard-code the search path to a set of known-safe values (such as system directories), or only allow them to be specified by the administrator in a configuration file. Do not allow these settings to be modified by an external party. Be careful to avoid related weaknesses such as CWE-426 and CWE-428.
- Implementation: When invoking other programs, specify those programs using fully-qualified pathnames. While this is an effective approach, code that uses fully-qualified pathnames might not be portable to other systems that do not use the same pathnames. The portability can be improved by locating the full-qualified paths in a centralized, easily-modifiable location within the source code, and having the code refer to these paths.
- Implementation: Remove or restrict all environment settings before invoking other programs. This includes the PATH environment variable, LD_LIBRARY_PATH, and other settings that identify the location of code libraries, and any application-specific search paths.
- Implementation: Check your search path before use and remove any elements that are likely to be unsafe, such as the current working directory or a temporary files directory. Since this is a denylist approach, it might not be a complete solution.
- Implementation: Use other functions that require explicit paths. Making use of any of the other readily available functions that require explicit paths is a safe way to avoid this problem. For example, system() in C does not require a full path since the shell can take care of finding the program using the PATH environment variable, while execl() and execv() require a full path.
References
Exploits and evidence
Timeline
- Published
- Last updated by the source
- Latest EPSS score
Related advisories
| Advisory | Severity | Why it's related |
|---|---|---|
| GHSA-8397-8qf6-f2gm vLLM: Phi-4 multimodal audio encoder materializes and retains full attention matrices: a single max-length audio OOM-crashes the vLLM engine | Medium | Also affects vllm |
| GHSA-x567-49g5-p4cf vLLM: Responses API bypasses trust_request_chat_template for chat_template_kwargs | Medium | Also affects vllm |
| GHSA-vvpg-rrww-wpmh vLLM: Permissive default CORS origins allow cross-origin requests to the vLLM HTTP server | Medium | Also affects vllm |
| GHSA-q97c-p368-75c4 vLLM: Out-of-vocabulary prompt token id passes input validation and triggers a CUDA device-side assert (engine crash / DoS) | Medium | Also affects vllm |
| GHSA-h3rc-j82p-2c3h vLLM: Untrusted EPS inputs can trigger denial of service or Ghostscript RCE | High | Also affects vllm |
| GHSA-gq5q-89c7-vhvx vLLM DeepSeek-OCR image processing vulnerable to denial of service | Medium | Also affects vllm |
| GHSA-6757-hggm-hp2w vLLM: Public multimodal SHM handle forgery reaches worker-side `pickle.loads()` with insecure serialization disabled | High | Also affects vllm |
| GHSA-542m-9234-8gvf vLLM: DoS via unbounded chat_template Jinja2 rendering exhausts the renderer thread pool | Medium | Also affects vllm |
MergeHarbor checks every repository's dependencies against this database, ranks what matters by exploitability and reachability, routes it to the owning team and opens the fix pull request.
Check your repos with MergeHarbor