UTC
Medium

vLLM: Flash late-interaction scoring caches query embeddings under a caller-controlled request id — cross-request integrity break and induce

CVE-2026-105755 · GHSA-2phq-3phc-84px · published 2026-10-05

Severity
Medium
CVSS score
4.2
Chance of exploitation (EPSS)
0.20%
Fixed in
0.30.0
  • Exploitation No known exploitation
  • Automatable No
  • Technical impact Partial
  • CISA SSVC
What it is

A network client of the standard scoring API can, on a flash late-interaction /score or /rerank deployment: 1. Corrupt another user's results — by reusing the victim's X-Request-Id, the attacker's query embedding overwrites the victim's cached entry, so the victim's documents are scored against the attacker's query (a cross-request integrity break). 2. Induce errors — depending on timing, one request consumes the shared use counter and forces the other request into a late-interaction cache-miss error. Both consequences follow deterministically from reusing the victim's header value, because same-key work is pinned to one engine. …

What to do

Upgrade vllm to 0.30.0 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.

Derive the flash late-interaction query-cache key from a server-generated, unforgeable per-request identifier (a random_uuid() namespace) rather than the caller-supplied X-Request-Id, and thread that key through the PoolingServeContext to the doc-scoring pass so both passes reuse the same key and a caller cannot address another request's cache entry. and the encode-docs pass reads those stored keys instead of re-deriving them from ctx.request_id: This requires adding the field to PoolingServeContext (vllm/entrypoints/pooling/typing.py). …

Affected
PackageRegistryAffectedFixed in
vllmPyPI< 0.30.00.30.0

The impact, patch and workaround text comes from the advisory, shortened (GitHub Advisory Database, CC BY 4.0). 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].