Description
Node-RED is an open-source, flow-based programming environment built on Node.js. It provides a browser-based editor that allows users to connect applications, devices, APIs and services through visual programming flows.
It is commonly used in IoT deployments, industrial automation, manufacturing, utilities, healthcare and other connected environments.
The vulnerability tracked as CVE-2026-58115 affects Siemens SIMATIC IoT2050 Advanced devices running Industrial OS with Node-RED installed.
The underlying issue is missing authentication on the Node-RED HTTP interface. This allows remote, unauthenticated attackers to access programming functionality capable of executing system commands.
VulnCheck reported that attackers were already attempting to exploit this weakness in the wild. Its canary infrastructure recorded exploitation attempts involving Linux payload delivery, command execution and alternative backdoor deployment methods.
Technical Description
CVE-2026-58115 – Missing Authentication for Critical Function
Affected product: Siemens SIMATIC IoT2050 Advanced (6ES7647-0BA00-1YA2)
Affected versions: All versions earlier than V4.3.4.1 running Industrial OS with Node-RED installed.
CWE: CWE-306 – Missing Authentication for Critical Function.
CVSS v3.1: 10.0 – Critical.
CVSS v4.0: 10.0 – Critical.
Attack vector: Network.
Privileges required: None.
User interaction: None.
The vulnerability allows attackers to access the Node-RED HTTP interface without authentication and create malicious flows.
Two built-in Node-RED node types are particularly relevant:
Inject node: Triggers a message manually, on a schedule or automatically when a flow starts.
Exec node: Executes system commands on the underlying server.
By connecting these nodes, an attacker can create a flow that automatically executes a command when deployed.
This enables remote command execution without requiring further interaction from the victim.
Attack Overview
Based on VulnCheck's observed exploitation activity, the attack sequence can be summarized as follows:
Target Identification: Attackers identify Node-RED instances exposed through the HTTP interface without authentication.
Vulnerability Exploitation: A malicious HTTP POST request is sent to the
/flowsendpoint.Malicious Flow Deployment: The request introduces a flow containing an Inject node and an Exec node.
Command Execution: The Inject node automatically triggers the Exec node after a short delay.
Command Obfuscation: A Base64-encoded shell command is decoded and executed through the Linux shell.
Payload Download: The command attempts to retrieve a Linux binary using
wget, withcurlas a fallback.Payload Execution: The downloaded file is made executable and launched as a detached process.
Alternative Backdoor Deployment: Subsequent requests were observed creating HTTP endpoints connected to command-execution nodes, allowing attackers to execute commands through later HTTP requests.
The observed exploitation attempts demonstrate how a single unauthenticated request can introduce a flow capable of executing operating-system commands.
Technical Analysis
1. Exploitation Through the Node-RED Admin API
The attacker sent an HTTP POST request to:
POST /flows
The request contained a Node-RED flow definition with three components:
An empty flow tab.
An Inject node configured to trigger automatically after a one-second delay.
An Exec node connected to the Inject node.
Once deployed, the flow automatically initiated command execution.
VulnCheck's canary returned an HTTP 200 response and a new flow revision hash, matching the expected response from a vulnerable Node-RED instance.
2. Multi-Layer Command Encoding
The attacker used three layers of encoding to conceal the execution process.
The first layer contained the malicious Node-RED flow definition.
The second layer included a shell command that decoded a Base64 string and passed its contents to the Linux shell.
The third layer contained the actual dropper commands.
This technique makes simple detection rules that search for strings such as wget or curl in the original HTTP request less effective.
3. Linux Payload Delivery
The decoded dropper attempted to download a Linux binary from an external server.
It used wget first and fell back to curl if the initial download failed.
The binary was saved at:
/dev/shm/.b
The /dev/shm directory is generally backed by temporary memory storage. This allows files to exist without being stored as ordinary persistent disk files.
The leading dot also makes the filename hidden from standard directory listings.
After downloading the file, the command applied executable permissions using chmod +x.
The process was then launched using setsid, with nohup as a fallback, allowing it to continue independently of the Node-RED process.
Important: VulnCheck did not retrieve or analyze the Linux binary. Its actual capabilities, including whether it functions as a botnet implant, cryptominer or another type of malware, remain unconfirmed.
4. Alternative Payload Delivery
In its November 1 update, VulnCheck reported additional payload variations.
One variant downloaded a shell script named wget.sh from the same payload server and executed it.
Another request attempted to retrieve a file named .sys_update from a different server and save it as:
/tmp/.kworker
The filename resembles the naming convention of legitimate Linux kernel worker processes.
The downloaded file was then executed using nohup.
The contents and functionality of these additional payloads were not recovered or analyzed.
5. Persistent Webshell Deployment
One of the most significant findings involved malicious Node-RED flows that did not download a separate binary.
Instead, these flows created an HTTP endpoint, such as /cmd, and connected it to a Function node and an Exec node.
The endpoint could receive commands through HTTP query parameters and pass them to the system command-execution node.
This effectively created a webshell-like mechanism that allowed attackers to execute commands remotely.
Unlike the initial memory-backed binary, this malicious flow could remain deployed even after a temporary payload disappeared following a reboot.
6. Attacker Infrastructure and Activity
VulnCheck identified multiple infrastructure indicators associated with the exploitation attempts.
The observed activity included:
16 events targeting CVE-2026-58115 from the same source IP between October 30 and November 1.
125 hits against eight different CVEs from that source IP across VulnCheck's wider canary telemetry.
Multiple user-agent strings, including curl, Python requests and Python urllib.
Several different payload delivery techniques.
The source IP was associated with Microsoft's AS8075 network and geolocated to Brazil.
However, the researchers cautioned that this does not identify the actual threat actor. The infrastructure may represent a compromised or rented cloud-hosted system used for opportunistic exploitation.
The observed activity is more consistent with broad opportunistic scanning and exploitation than a confirmed targeted campaign.
Indicators of Compromise (IOCs)
The following indicators were published in VulnCheck's investigation.
Network Indicators
| Type | Indicator | Description |
|---|---|---|
| IP | 4.201.105.116 | Observed exploitation source |
| IP | 85.215.73.250 | Linux payload server |
| IP | 45.194.90.57 | Alternative payload server |
| URL | http://85.215.73[.]250:8080/bot-linux-amd64 | Linux binary download |
| URL | http://85.215.73[.]250:8080/wget.sh | Staged shell-script download |
| URL | http://45.194.90[.]57:38472/.sys_update | Alternative payload download |
File and Path Indicators
| Indicator | Description |
|---|---|
/dev/shm/.b | Memory-backed Linux payload path |
/tmp/.s | Staged shell script path |
/tmp/.kworker | Alternative payload path |
wget.sh | Staged download script |
.sys_update | Alternative downloaded payload |
Other Observed Artifacts
Node-RED flow deployment through
POST /flows.Automatically triggered Inject nodes connected to Exec nodes.
Base64-encoded shell commands.
Unexpected HTTP endpoints connected to command-execution functionality.
IOC handling note: These indicators are reproduced from VulnCheck's research. Validate them against current telemetry and threat-intelligence feeds before deploying blocking or detection rules.
Potential Impact
Successful exploitation of CVE-2026-58115 can provide an attacker with arbitrary command execution on an affected device with maximum privileges.
Potential consequences include:
System Compromise: Unauthorized command execution can provide extensive control over vulnerable devices.
Malware Deployment: Attackers may install additional malicious binaries or scripts.
Persistent Unauthorized Access: Malicious Node-RED flows can provide remote command-execution functionality.
Network Exposure: Compromised IoT or industrial devices may provide attackers with a foothold inside connected networks.
Operational Disruption: Unauthorized changes to industrial automation environments could affect connected services and operational processes.
Data Exposure: Attackers with sufficient access may attempt to collect sensitive system or application information.
The potential impact is particularly important in OT environments, where compromised connected devices may affect the availability and integrity of industrial operations.
These are potential consequences of successful compromise, not confirmed outcomes for every observed exploitation event.
Recommendations
Immediate Actions
Identify Siemens SIMATIC IoT2050 Advanced devices running Industrial OS with Node-RED installed.
Verify the firmware version and update affected devices to V4.3.4.1 or later.
Restrict external and unnecessary network access to the Node-RED HTTP interface.
Review deployed Node-RED flows for unexpected Inject, Exec, Function and HTTP In nodes.
Investigate suspicious flow deployments and unauthorized changes to flow revisions.
Search Linux systems for unexpected files in
/dev/shmand/tmp, including.b,.sand.kworker.Investigate suspicious
wget,curl,chmod,setsidandnohupexecution chains.Review outbound network connections involving the published IP addresses and URLs.
If compromise is suspected, isolate affected devices according to OT incident-response procedures and preserve forensic evidence.
Preventive Measures
Enable authentication for the Node-RED editor and administrative API.
Follow Node-RED security-hardening guidance and use HTTPS for access.
Avoid exposing Node-RED management interfaces directly to the public internet.
Place IoT and OT devices behind appropriate network segmentation and access controls.
Apply least-privilege principles to Node-RED processes and service accounts.
Monitor flow creation, modification and deployment activities.
Implement network monitoring to detect unexpected outbound connections and command execution.
Maintain updated firmware and security patches for embedded industrial products.
Establish continuous vulnerability assessment and threat-hunting procedures for exposed IoT and OT infrastructure.
Conclusion
The exploitation of CVE-2026-58115 demonstrates how an exposed and unauthenticated Node-RED interface can become a direct entry point for remote command execution.
VulnCheck's observations show that attackers can deploy malicious flows, execute encoded shell commands, retrieve Linux payloads and establish webshell-like command-execution endpoints through the same vulnerable interface.
The incident is particularly relevant to organizations using Node-RED within IoT and industrial environments, where a compromised device may introduce risks beyond an individual endpoint.
Organizations should prioritize updating affected Siemens devices, securing Node-RED administrative interfaces and monitoring unexpected flow deployments and command-execution activity.
The incident also reinforces an important security principle: development and automation interfaces must be treated as privileged administrative surfaces, not ordinary application endpoints.
