惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

雷峰网
雷峰网
B
Blog
博客园_首页
云风的 BLOG
云风的 BLOG
S
SegmentFault 最新的问题
罗磊的独立博客
Jina AI
Jina AI
C
Check Point Blog
Martin Fowler
Martin Fowler
J
Java Code Geeks
博客园 - 司徒正美
美团技术团队
MongoDB | Blog
MongoDB | Blog
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
大猫的无限游戏
大猫的无限游戏
有赞技术团队
有赞技术团队
U
Unit 42
Hugging Face - Blog
Hugging Face - Blog
WordPress大学
WordPress大学
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
博客园 - 叶小钗
博客园 - 三生石上(FineUI控件)
小众软件
小众软件

Security Affairs

Digital attacks drive a new wave of cargo theft, FBI says Carding service Jerry’s Store leak exposes 345,000 stolen payment cards Anthropic launches Claude Security to counter rapid AI-Powered exploits SonicWall patches three SonicOS flaws in Gen 6, 7 and 8 firewalls. Patch them now Copy Fail: New Linux bug enables Root via page‑cache corruption Agent’s claims on WhatsApp access spark security concerns Meta accused of violating DSA by failing to safeguard minors Large-scale Roblox hacking operation shut down by Ukrainian authorities CVE-2026-42208: LiteLLM bug exploited 36 hours after its disclosure Internet censorship index reveals Russia’s lead and widespread content blocking All supported cPanel versions hit by critical auth bug, now patched U.S. CISA adds Microsoft Windows Shell and ConnectWise ScreenConnect flaws to its Known Exploited Vulnerabilities catalog ShinyHunters exploit Anodot incident to target Vimeo CVE-2026-3854 GitHub flaw enables remote code execution Signal Phishing Campaign Targets German Officials in Suspected Russian Operation Microsoft fixes Entra ID flaw enabling privilege escalation New Android spyware Morpheus linked to Italian surveillance firm NCSC launches SilentGlass, a plug-in device to secure HDMI and DisplayPort links Medtronic discloses security incident after ShinyHunters claimed theft of 9M+ records Chinese spy posed as researcher in spear-phishing campaign targeting NASA to steal defense software LINKEDIN BROWSERGATE Firefox bug CVE-2026-6770 enabled cross-site tracking and Tor fingerprinting Fast16: Pre-Stuxnet malware that targeted precision engineering software Italy moves to extradite Chinese national to the U.S. over hacking charges U.S. utility giant Itron discloses a security breach Critical bug in CrowdStrike LogScale let attackers access files GopherWhisper: new China-linked APT targets Mongolia with Go-based malware SECURITY AFFAIRS MALWARE NEWSLETTER ROUND 94 Trigona ransomware adopts custom tool to steal data and evade detection Security Affairs newsletter Round 574 by Pierluigi Paganini – INTERNATIONAL EDITION
CVE-2026-33032: severe nginx-ui bug grants unauthenticate...
2026-04-16 · via Security Affairs

An actively exploited critical nginx-ui flaw (CVE-2026-33032) lets attackers bypass authentication and take full control of Nginx servers.

A critical vulnerability in nginx-ui, tracked as CVE-2026-33032 (CVSS score of 9.8), is being actively exploited, allowing attackers to bypass authentication and fully take over Nginx servers. The issue stems from improper protection of the /mcp_message endpoint, which relies only on IP whitelisting. Since the default whitelist allows all, attackers can access the service without authentication and gain control.

“The nginx-ui MCP (Model Context Protocol) integration exposes two HTTP endpoints: /mcp and /mcp_message. While /mcp requires both IP whitelisting and authentication (AuthRequired() middleware), the /mcp_message endpoint only applies IP whitelisting – and the default IP whitelist is empty, which the middleware treats as “allow all”.” reads the advisory. “This means any network attacker can invoke all MCP tools without authentication, including restarting nginx, creating/modifying/deleting nginx configuration files, and triggering automatic config reloads – achieving complete nginx service takeover.”

Yotam Perkal of Pluto Security discovered the nginx-ui flaw. The researcher pointed out that it can be exploited in seconds using just two HTTP requests.

“The attack flow: step 1 authenticates to get a session, step 2 uses that session to invoke destructive tools with zero authentication.” wrote Perkal.An attacker on the same network as the nginx-ui instance needs just two requests:

  • POST /mcp_message?sessionId=xxx – Invoke any tool. No node_secret. No JWT. No cookies. Nothing.
  • GET /mcp?node_secret=xxx – Establish an SSE session, get a sessionId”

A single unauthenticated request can let attackers fully compromise nginx-ui. They can intercept all traffic by redirecting it through malicious servers, capture admin credentials via manipulated logs, and gain persistent access by stealing tokens and secrets. Attackers can also map the entire infrastructure by reading configs and even shut down services by pushing invalid configurations.

The flaw was fixed in nginx-ui version 2.3.4 by adding a missing authentication check to the /mcp_message endpoint, just one line of code. The update also introduced a regression test to ensure both endpoints require authentication, which would have prevented the issue. Notably, some version trackers are incorrect: v2.3.3 is the last vulnerable release, while v2.3.4 includes the fix.

Follow me on Twitter: @securityaffairs and Facebook and Mastodon

Pierluigi Paganini

(SecurityAffairs – hacking, nginx)