Know every vulnerabilitybefore it knows you.
DevGuard continuously monitors your dependencies and alerts you when CVEs like this one affect your stack — with real-time threat intelligence built for developers.
GHSA-h3rc-6mm3-gc2m
Summary
An untrusted multimodal request can set mm_processor_kwargs.code_revision. When vLLM serves a model whose Hugging Face processor uses dynamic auto_map code and the operator enabled --trust-remote-code, this request value reaches AutoProcessor.from_pretrained and selects the Python revision imported by Transformers.
This lets a client replace the operator-reviewed processor module with code from another revision of the same model repository. The operator-selected model/data revision remains pinned; only the executed processor code changes.
Importing a Python module is arbitrary code execution in the vLLM API/renderer process. The PoC below demonstrates this with an import-time filesystem write. It is fully offline and does not require a GPU, model weights, or changes to a real Hub repository.
Affected snapshot
- Repository:
vllm-project/vllm - Verified commit:
570735520942b6428ae98f1eb442c6d88adae55e - Verified with Transformers 5.15.0 and 5.16.1
- Affected release range has not been bisected
Affected configuration
The vulnerable path exists when all of the following are true:
- The served model's processor uses dynamic
auto_map.AutoProcessorcode. - The operator enabled
trust_remote_code. - The server can fetch the request-selected revision, or it is already cached.
- The attacker can create or otherwise control a revision in the served model repository. An example is a Hugging Face community pull-request ref when that repository permits them.
- A multimodal request reaches the processor path.
The default trust_remote_code=False configuration is not affected.
Attack surface
mm_processor_kwargs is accepted directly by public request models:
/tokenize is particularly relevant because vLLM's authentication middleware guards only /v1, /v2, /inference, and /cohere prefixes:
Consequently, /tokenize remains unauthenticated even when --api-key is configured, unless an external proxy protects it. The vLLM Security Guide documents /tokenize as an unprotected endpoint.
Root cause
The request value follows this path:
OnlineRenderer.preprocess_chatplaces requestmm_processor_kwargsin prompt extras.- Multimodal rendering forwards those kwargs into processor construction through
MMProcessorInputsand_process_multimodal_async. MultiModalConfig.merge_mm_processor_kwargsoverlays inference values on startup values, so the request wins.cached_get_processor_without_dynamic_kwargsfilters only recognized processor-call keys.code_revisionis retained as a processor-construction kwarg.cached_processor_from_configpinsrevisionandtrust_remote_code, but forwards the remaining kwargs without reservingcode_revision.get_processorcallsAutoProcessor.from_pretrained(..., revision=revision, trust_remote_code=trust_remote_code, **kwargs).- Transformers
AutoProcessor.from_pretrainedusescode_revisionto retrieve and import the dynamic module throughget_class_from_dynamic_module, then removes it before loading processor data. This is why code comes from the attacker revision while configuration/data remain operator-pinned.
Request-time processor arguments are not separated from processor-construction arguments, and code_revision is not reserved.
Suggested remediation
- Separate processor-construction arguments from untrusted processor-call arguments. An allowlist of supported request-time preprocessing controls is safest.
- Reject
code_revisionand other loader/source controls in requestmm_processor_kwargs, and passmodel_config.code_revisionexplicitly as the only permitted processor-code revision. - Add a regression test proving a request cannot change the loaded module revision while operator pins remain effective.
- Protect
/tokenizewith the configured authentication policy as defense in depth.
Expected safe behavior is either a 4xx response for reserved request keys or silently ignoring them before processor construction. Rejection is preferable because it makes attempted misuse visible.
A patch for this was merged here: https://github.com/vllm-project/vllm/pull/58830
Upload your own SBOM in CycloneDX 1.6 or higher (JSON) directly here to check your vulnerabilities.
Drag and drop some file here, or click to select
The vulnerability can be exploited over the network without needing physical access. It is difficult for an attacker to exploit this vulnerability and may require special conditions. An attacker does not need any special privileges or access rights. No user interaction is needed for the attacker to exploit this vulnerability. The impact is confined to the system where the vulnerability exists. There is a high impact on the confidentiality of the information. There is a high impact on the integrity of the data. There is a high impact on the availability of the system.
Exploitation activity has been observed. Apply available patches or mitigations urgently.
The exploit probability is very low. The vulnerability is unlikely to be exploited in the next 30 days.
We did not find any exploit available. Neither in GitHub repositories nor in the Exploit-Database.
Browse More
Continuously monitor your dependencies and get alerted when vulnerabilities like this one affect your stack.
Checkout DevGuard