# N-central Hotfix 3 ships 17 fixes and ends the Legacy API

_November 17, 2026 · LNX Inc._

N-able ships N-central HF3 with 17 fixes; Barracuda traces its breach to an unretired vendor key; Optimizely scripts turned skimmer. One pattern links them.

N-able has shipped the N-central fix that the [November 15 briefing](/news/msp-threat-briefing-2026-11-15/) was still waiting for. It is one of four stories this Tuesday, November 17, 2026, and all four have the same shape. An attacker did not break the victim directly. They went through something the victim had trusted with access: an RMM agent, a remote-support extension, a vendor's cloud key, a marketing script.

## Update: Hotfix 3 closes the N-central door for good

N-able released [N-central 2026.4 Hotfix 3](https://status.n-able.com/2026/11/16/n-central-2026-4-hf3-release-notes/), build 2026.4.1.39, on November 16. It fixes 17 CVEs, CVE-2026-129271 through CVE-2026-129287. They include missing authentication on server endpoints, XML external entity injection, SSRF and a path traversal that leads to code execution. The release notes call it a critical security update. They do not say any of the 17 are exploited.

The bigger change is structural. The Legacy API option is gone and the legacy APIs are permanently blocked. That is the setting the 4EverAble2Shell and Agent2Shell chains needed, the ones used to [mine Monero on Hotfix 2 servers](/news/msp-threat-briefing-2026-11-13/). Remote control other than Take Control stays disabled, and the Intune integration stays blocked.

The line that matters most for the field is about agents. N-able says agents and probes that are not upgraded are "susceptible to impersonation." Patching the server is half the job. Each managed device also needs the HF3 agent. After upgrading, N-central now rejects devices without an SMBIOS ID or machine ID, and a new Inactive Assets tab lists agents that have stopped reporting. Hosted N-central customers get the server upgrade automatically, but not the agent work.

## The RMM agent and the support extension are now the perimeter

ConnectWise said on November 16 at 20:00 UTC that it has [completed its investigation](https://www.connectwise.com/company/trust/advisories) into the ScreenConnect activity. It adds that completion "does not establish" that any partner environment is unaffected. [Huntress](https://www.huntress.com/blog/sckit-worm-rogue-screenconnect) tied the attacks to the third-party RMM+ extension. Versions 1.0.31 and 1.0.32 accepted a blank access key and bypassed ScreenConnect's own accounts, roles and MFA. A second wave on November 13 reached 205 hosts across 41 organizations.

The toolkit, SCKit, includes a relay trap built to spread to the technician who connects to an infected machine. Huntress reconstructed that path from code but did not see it work. Still, the guidance is to investigate suspect hosts with WinRM or another out-of-band tool, not with ScreenConnect.

| Indicator (from Huntress) | Meaning |
|---|---|
| `185.241.211[.]197` | Host for the rogue client, November 13 wave |
| Service `Windows Update Agent` running as LocalSystem | Rogue client persistence |
| `C:\ProgramData\Windows Update Agent (de362a425905bcc4)\UpdateConnect.ClientService.exe` | Rogue client binary |
| Scheduled task `TrapWatch`, folder `C:\ProgramData\RelayTrap\` | Relay trap persistence |
| Run value `RelayTrapLiteAlert` under `HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Run` | Relay trap persistence |

N-central and ScreenConnect differ in every technical detail. The lesson is the same. A tool that can run code on every client machine is only as safe as its weakest plug-in, API mode or out-of-date agent.

## A key nobody retired opened Barracuda's cloud

Barracuda published a [frank account](https://blog.barracuda.com/2026/11/11/anatomy-of-a-machine-speed-intrusion) of its own breach, and BleepingComputer [reported on November 16](https://www.bleepingcomputer.com/news/security/barracuda-links-cloud-breach-to-artex-ai-powered-cyberattack/) that it links the intrusion to ARTEX. Barracuda had given a cloud access key to an AI vendor for a past project and never revoked it. The vendor's development environment was compromised in October. From 04:33 UTC on November 1 to 23:31 UTC on November 2, an attacker used the key to touch about 10 cloud accounts.

Barracuda found no direct evidence that the credentials its XDR service uses to collect customer data were accessed. Its logging could not rule it out. The [Trust Center notice](https://trust.barracuda.com/security/barracuda-xdr-integration-credentials) asks notified Managed XDR Cloud and Monitored Endpoint customers to revoke and recreate those integration credentials. They should also review activity in the connected systems, such as Microsoft 365 and Google Workspace, from November 1. Customers Barracuda did not contact need take no action.

Every MSP has keys like that one: API tokens handed to a reporting tool, an app registration for a migration that ended last year. ARTEX is the [kit the FBI flagged on Monday](/news/msp-threat-briefing-2026-11-16/). It is built to find exactly those forgotten credentials.

## A trusted script tag became a card skimmer

On November 16, someone wrote malicious code into customer projects on Optimizely Web Experimentation. The code was then served from Optimizely's own CDN. Researcher Scott Helme [checked 1,611 Optimizely snippets](https://blog.report-uri.com/worlds-largest-brands-infected-with-malware-via-optimizely/) and found 188, belonging to 178 organizations, infected. The payload swapped crypto wallet addresses and skimmed cards typed into site forms. It also sent some cookies and screenshots to `1688360[.]cc` and loaded iOS exploit kits on iPhones and iPads. Optimizely's [status page](https://status.optimizely.com) says the vulnerability is fixed and the injected code removed.

Few small businesses run Optimizely. Many run some third-party script that can change without their knowledge: chat widgets, analytics, A/B testing, review badges. Helme found that a restrictive Content Security Policy `script-src` stopped the payload on some affected sites, while many had no policy at all.

## Speed removes the grace period

Google's Threat Intelligence Group [reported on November 16](https://cloud.google.com/blog/topics/threat-intelligence/agentic-threat-activity-crosses-a-significant-threshold) that attackers now use language models to diff patches and build exploits automatically. With the PaperCut flaws in August, researchers had working exploits within a day of disclosure. GreyNoise saw a campaign four days after it. N-central's 4EverAble2Shell chain was [exploited within hours](https://www.bleepingcomputer.com/news/security/new-n-able-4everable2shell-zero-day-exploited-hours-after-poc-released/) of its proof of concept going public, as reported by BleepingComputer. A hotfix that sits in a change window for two weeks may now be past its useful life when it lands.

## Do this

Upgrade on-premises N-central servers to build 2026.4.1.39 today, then push the HF3 agent and probe to every device and work the Inactive Assets list. On ScreenConnect, confirm the RMM+ family stays disabled, technician clients are on 26.6.5 or later, and no host shows the indicators above. If Barracuda notified you, rotate the XDR integration credentials and review connected tenants from November 1. Then do the slower job that all four stories point to. List every API key, app registration and integration token that reaches into your clients, and revoke anything without a current owner. For client websites, inventory third-party scripts and put a restrictive Content Security Policy in front of them.

## Sources

- [N-able: N-central 2026.4 HF3 release notes](https://status.n-able.com/2026/11/16/n-central-2026-4-hf3-release-notes/)
- [The Hacker News: N-able patches 17 N-central flaws as attackers exploit earlier bugs to deploy crypto miners](https://thehackernews.com/2026/11/n-able-patches-17-n-central-flaws-as.html)
- [ConnectWise: Trust Center advisories (ScreenConnect)](https://www.connectwise.com/company/trust/advisories)
- [Huntress: SCKit worm and rogue ScreenConnect clients](https://www.huntress.com/blog/sckit-worm-rogue-screenconnect)
- [The Hacker News: ConnectWise warns ScreenConnect attacks continue as SCKit targets unpatched systems](https://thehackernews.com/2026/11/connectwise-warns-screenconnect-attacks.html)
- [Barracuda: Anatomy of a machine-speed intrusion](https://blog.barracuda.com/2026/11/11/anatomy-of-a-machine-speed-intrusion)
- [Barracuda: XDR integration credentials advisory](https://trust.barracuda.com/security/barracuda-xdr-integration-credentials)
- [BleepingComputer: Barracuda links cloud breach to ARTEX AI-powered cyberattack](https://www.bleepingcomputer.com/news/security/barracuda-links-cloud-breach-to-artex-ai-powered-cyberattack/)
- [Report URI: World's largest brands infected with malware via Optimizely](https://blog.report-uri.com/worlds-largest-brands-infected-with-malware-via-optimizely/)
- [Optimizely: status page](https://status.optimizely.com)
- [Google Threat Intelligence Group: Agentic threat activity crosses a significant threshold](https://cloud.google.com/blog/topics/threat-intelligence/agentic-threat-activity-crosses-a-significant-threshold)
