VulnCheck reports that attackers began exploiting CVE-2026-58115, a maximum-severity (CVSS 10.0) missing-authentication flaw in Node-RED, on 30 October 2026, using a single HTTP request to run commands on exposed servers. According to Help Net Security and SecurityWeek, the first hit landed on a VulnCheck decoy system, and the same source address tried the flaw 16 times by 1 November. Neither report gives a count of exposed instances.
Node-RED is a low-code, flow-based tool for building event-driven applications. It is also embedded in commercial industrial products. Siemens disclosed CVE-2026-58115 in August 2026 (advisory SSA-834709): an unauthenticated remote attacker can create malicious flows and execute arbitrary code on the server with maximum privileges. Siemens lists the SIMATIC IoT2050 Advanced running Industrial OS with Node-RED as affected, fixed in V4.3.4.1 and later. The reports do not name affected or fixed upstream Node-RED releases.
Per VulnCheck's observations:
/flows deployed a flow with three layers of encoding. An inject node triggered an exec node automatically, so the shell command ran on deployment.bot-linux-amd64 from a separate server and ran it in the background from a memory-backed location. VulnCheck did not retrieve the binary, so its purpose is unknown.The attacks look for Node-RED instances that have adminAuth disabled. CISA had not listed the flaw in its KEV catalog at the time of VulnCheck's report; VulnCheck's own KEV catalog has it.
adminAuth enabled on every Node-RED instance you run?The advisory was public for nearly three months before exploitation started. The gap that matters is not the patch, it is knowing where the software is. Node-RED rarely appears in a procurement record as "Node-RED"; it arrives inside an industrial gateway, an integration box a contractor installed, or a developer's side project. Teams that track servers and laptops can still have no list of what these devices expose.
Obiguard Scout approaches this from the outside, the way the scanners do. It enumerates the hosts and services your organisation publishes to the internet and records what it observes, so an unexpected Node-RED editor port on a gateway, or an HTTP interface that answers without authentication, shows up in your exposure list rather than in an incident report. That gives you the first thing the checklist above needs: the set of reachable candidates to verify. Scout sees only what is visible from outside; confirming adminAuth settings and reviewing flows still requires access to the device. For a similar "known but unfound" exposure story, see our GitLab and Alpine analysis. You can run a free Scout check on a domain you own to see what is already published under it.
Obiguard sits in front of every AI request your organization makes — screening prompts and outputs against the guardrails, compliance frameworks, and audit trails that stories like this one make necessary.
See how it works →