UTC
High

Remote DoS in vLLM via allowed_token_ids Speculative-Decoding Metadata Mismatch

GHSA-8c65-hq7q-r7jm · published 2026-10-09

Severity
High
CVSS score
7.5
Chance of exploitation (EPSS)
No CVE
Fixed in
0.18.0
What it is

One accepted third request can crash the engine worker when two earlier requests are still in flight In serve mode, in-flight requests are aborted and later requests fail with either 500 or connection failure after the worker dies In shared deployments, a single request can take the endpoint offline for all clients Repeating the request can keep the service down

What to do

Upgrade vllm to 0.18.0 or later. If you start it from an MCP config with npx or uvx, pin that version so an old cached copy is not used.

Affected
PackageRegistryAffectedFixed in
vllmPyPI>= 0.11.1, < 0.18.00.18.0

The impact, patch and workaround text comes from the advisory, shortened (published by vLLM). Proof-of-concept and reproduction sections and code are left out; the original advisory has them.

Sources: GitHub Advisory Database, reviewed and malware advisories (CC BY 4.0, https://github.com/advisories); advisories published by Cursor (https://github.com/cursor/cursor/security/advisories); advisories published by OpenClaw (https://github.com/openclaw/openclaw/security/advisories); advisories published by vLLM (https://github.com/vllm-project/vllm/security/advisories); advisories published by Open WebUI (https://github.com/open-webui/open-webui/security/advisories); advisories published by n8n (https://github.com/n8n-io/n8n/security/advisories); advisories published by Flowise (https://github.com/FlowiseAI/Flowise/security/advisories); advisories published by Langflow (https://github.com/langflow-ai/langflow/security/advisories); advisories published by the MCP Python SDK project (https://github.com/modelcontextprotocol/python-sdk/security/advisories); advisories published by MLflow (https://github.com/mlflow/mlflow/security/advisories); CISA Known Exploited Vulnerabilities catalog (CC0). Data as of 2026-11-22. ShieldMap describes how each problem works and how to fix it; it does not publish exploit code. Report a correction to [email protected].