---
id: "CVE-2026-111391"
kind: "vulnerability"
aliases: ["CVE-2026-111391"]
title: "SGLang 0.5.10 through 0.5.22 contains a code injection vulnerability in _evaluate_filter() that allows remote\u2026"
severity: "critical"
cvss: 9.2
epss: null
exploited: false
status: "open"
category: "ai-tooling"
products: []
ecosystems: []
tags: ["ai-tooling", "llm-serving", "python", "self-hosted", "sglang"]
not_affected_if: ["default-config", "versions-outside-0.5.10-0.5.22"]
published_at: "2026-10-16T12:16:33+00:00"
first_seen_at: "2026-10-23T21:57:12+00:00"
updated_at: "2026-10-23T22:28:55+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-111391", "first_seen_at": "2026-10-23T21:57:12+00:00"}]
---

# SGLang 0.5.10 through 0.5.22 contains a code injection vulnerability in _evaluate_filter() that allows remote…

SGLang, an LLM serving framework, has a code injection flaw in its dumper filter evaluation that lets remote attackers run Python code. An attacker escapes the eval() sandbox through re.search __globals__ and sends a crafted filter to an unauthenticated endpoint.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

Check whether the environment variable DUMPER_SERVER_PORT is set on SGLang servers running versions 0.5.10 through 0.5.22.

## Fix

Upgrade SGLang to a version newer than 0.5.22 if one is released, or unset DUMPER_SERVER_PORT and block network access to the dumper port.

## Timeline

- 2026-10-23T21:57:12+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-111391

## From the sources

SGLang 0.5.10 through 0.5.22 contains a code injection vulnerability in _evaluate_filter() that allows remote attackers to execute Python code by escaping the eval() sandbox through re.search __globals__. Attackers can send a crafted filter expression to the unauthenticated /dumper/configure endpoint when DUMPER_SERVER_PORT is set, running code in the scheduler process on the next forward pass.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_