CRITICAL: N-able N-central Hotfix 3 Fixes 17 Flaws After Miner Attacks
N-able released N-central 2026.4 Hotfix 3 with fixes for 17 CVEs, led by the CVSS 9.3 appliance impersonation flaw CVE-2026-129272. The update permanently removes the Legacy API that attackers abused to plant XMRig miners, and every agent and probe must be upgraded to block impersonation.
If you run N-central on premises, last week ended with two pieces of news. Attackers were already planting cryptocurrency miners on N-central servers, and many of those servers had installed the hotfix N-able said would protect them. N-able's answer is N-central 2026.4 Hotfix 3, released on November 16, 2026. It fixes 17 more CVEs and removes the Legacy API from the product for good. It also changes how agents and probes authenticate to the server. Read the release notes carefully, because this one comes with homework on every endpoint, not just the server.
The most severe CVE in the batch is CVE-2026-129272, which N-able scored at 9.3 on CVSS 4.0. N-central releases before 2026.4.1 HF3 authenticate appliances to an agent-facing API with a credential an attacker can predict. A remote attacker with no credentials who knows or guesses an appliance identifier can get a valid appliance token and session, and then impersonate that appliance. Think about what that means on an RMM platform. The server trusts the appliances and agents that report in, and an attacker who can pose as one is talking to the system that pushes scripts and patches to every managed endpoint an MSP has.
The Week That Led Here
Some context explains why N-able is now on its third hotfix in a week. Over the weekend of November 8, independent researcher Sean Bales published proof-of-concept code for what he says are 33 N-central flaws, under the name Able2Shell. He went public without coordinating with the vendor. He has said that was because N-able had not fixed two unrelated bugs he reported back in July. N-able released Hotfix 2, build 2026.4.1.26, on November 10. It covered seven CVEs, the most serious being an unauthenticated XXE chain tracked as CVE-2026-125883 and an SSRF flaw tracked as CVE-2026-125881, both scored 9.3.
Two days later, Bales published new chains he called 4EverAble2Shell and Agent2Shell, and he claimed they worked against Hotfix 2. The first gives an attacker root on the N-central server. The second reaches SYSTEM on managed Windows agents. On November 12, the security firm Huntback reported that the 4EverAble2Shell chain had been used to install XMRig Monero miners on 21 live N-central servers. Nineteen of those servers reported that they were running Hotfix 2. We covered that chain and its Apache Axis root cause, CVE-2026-90436, last week. Both 4EverAble2Shell and Agent2Shell require the Legacy API mode to be enabled. N-able says the mode is off by default, but on at least some of the servers that matter, someone had turned it on.
What Hotfix 3 Actually Fixes
The 17 CVEs run from CVE-2026-129271 to CVE-2026-129287. Most of them follow one theme. N-central had too many doors with no lock on them, and too many rooms where a low-privilege user could reach things they had no business touching.
Several bugs in the ServerMMS and ServerMMS2 interfaces let a remote caller do things without logging in. CVE-2026-129274 and CVE-2026-129276 allowed task results and asset data to be submitted with no authentication at all. CVE-2026-129277 exposed reads of maintenance actions, and CVE-2026-129275 let the repositoryItemGet call disclose policies to anyone who asked. CVE-2026-129278 let an attacker tamper with Take Control audit logs without authenticating. That is the kind of bug you're glad to have patched before you ever need those logs during an incident.
Code execution shows up a few times. CVE-2026-129286 has a CVSS 4.0 score of 9.2. N-central did not validate version values and download locations in software update metadata, so an attacker who could supply or modify that metadata could write files outside the download directory and execute code as the N-central service account. The attack complexity is high and some preconditions apply, which is why it scores just under the appliance bug. CVE-2026-129273 is a syslog injection in the Audit Export setting. A product administrator could slip control characters into the system logging configuration and run commands as root. NVD lists it at 8.6 because it needs high privileges. Still, root on the appliance is a bigger prize than admin in the web console, and that gap is exactly what attackers chain toward. CVE-2026-129271 adds a local privilege escalation in Patch Management. CVE-2026-129282 is an argument injection in the test email sendmail function, with missing authorization checks on top.
The remaining fixes cover XML and request forgery bugs, which have been the theme of this whole saga. CVE-2026-129283 and CVE-2026-129287 are XXE injections, one in custom service import and one in the SIS repository parser. CVE-2026-129284 is an SSRF in Report Manager and the Data Export settings. CVE-2026-129280 covers missing authorization on ServerUI SOAP operations, and CVE-2026-129281 lets view-only users set DeviceExecute configuration, which is roughly as bad as it sounds. CVE-2026-129279 enables user enumeration through the password reset response. CVE-2026-129285 is an origin validation error in XMPP action responses.
N-able's own release notes don't say that any of these bugs has been exploited, and the vendor posted no severity ratings at release time. The scores mentioned above come from the CVE records that have since been published to NVD. Bales has also published details for 24 CVEs on GitHub, with the note that "these are patched vulnerabilities and should not work on the latest patch." Anyone who has watched this story for a week will take "should not" with a grain of salt. Still, once a public writeup exists for each bug, reverse engineering the fix becomes a weekend project rather than a research effort. Google's own threat intelligence team reported this week that attackers are using AI agents to build working exploits directly from software patches. The window between a fix and a working exploit is shrinking, and it wasn't large to begin with.
Breaking Changes You Need to Plan For
Hotfix 3 isn't a drop-in update, and there are a few things you should know before you click upgrade. The Legacy API configuration option has been removed completely. Any integration, script, or PSA connector that still depends on it will stop working, and there is no switch to turn it back on. That's the right decision, since the Legacy API is the precondition for the exploited chains. Some MSPs will still find out which of their homegrown automation was quietly using it.
Remote control and the Intune integration remain blocked for now. That continues the restrictions N-able introduced in Hotfix 2 and in the days before this release. Take Control was still functional after Hotfix 2, but plan for technician workflows to feel clumsy for a while.
The biggest operational item is the agent and probe change. N-able says that agents and probes that aren't upgraded "are susceptible to impersonation," and that customers "must upgrade all of their Agents and Probes to the latest version." That applies to hosted and on-premises customers alike. Hosted customers get the server update automatically, but the endpoints are still your job. Keep the ordering guidance from the Hotfix 2 rollout in mind as well. Upgrade agents first and probes second, because an upgraded probe can't talk to agents that haven't been upgraded. If you get the order wrong, you'll spend the afternoon working out why half a client site went dark.
What to Do Today
On-premises customers should treat this as an emergency change and upgrade the N-central server to 2026.4 Hotfix 3 now. If you can't patch right away, follow N-able's earlier guidance and keep unpatched servers off the internet until you can. Then push the updated agents and probes across every customer, in the order described above, and track completion for each site rather than assuming the rollout finished.
Because miners have already been found on live servers, patching isn't enough. You also need to hunt. Review N-central activity from November 8 onward. Look for remote access sessions nobody can explain, new or modified scripts and scheduled tasks, administrator accounts that showed up this month, and any change to the Legacy API setting in the server configuration. On the appliance, look for unexpected processes with sustained high CPU use, which is the classic XMRig signature, and for outbound connections to mining pools. Check the host artifacts Huntback published for the miner activity and the network indicators N-able released on November 13. If an attacker had root on your N-central server, assume they could also reach your managed endpoints. In that case, rotate the credentials stored in N-central, including the domain and local admin accounts your automation uses to log in.
Finally, check what's exposed to the internet. If your N-central server is reachable from anywhere, limit access to the agent-facing interfaces and the admin console with an allowlist or a VPN. That limits the damage from whatever turns up in next week's bug reports.
If an IT provider manages your systems
Many IT providers use N-central to push scripts and patches to every computer they manage, including yours. Attackers have already planted cryptocurrency miners on 21 live N-central servers, so ask your provider whether their system was one of them.
- Do you use N-central, and has your server been upgraded to 2026.4 Hotfix 3?
- Have the N-central agents on our computers been upgraded to the latest version, and how did you confirm every one finished?
- Did you check N-central activity since November 8 for signs of compromise and rotate the passwords stored in it?
A good answer is specific: dates, versions and what they checked, not just "we're fine." If your provider can't say whether their server was compromised, or finds remote sessions, scripts or admin accounts nobody can explain, bring in incident response to check your own systems.
If you are an IT provider yourself, see how we partner with MSPs.
References
- NVD CVE-2026-129272
https://nvd.nist.gov/vuln/detail/CVE-2026-129272
- NVD CVE-2026-129286
https://nvd.nist.gov/vuln/detail/CVE-2026-129286
- NVD CVE-2026-129273
https://nvd.nist.gov/vuln/detail/CVE-2026-129273
- The Hacker News: N-able Patches 17 N-central Flaws as Attackers Exploit Earlier Bugs
https://thehackernews.com/2026/11/n-able-patches-17-n-central-flaws-as.html
- Tenable Able2Shell FAQ
https://www.tenable.com/blog/able2shell-frequently-asked-questions-for-reported-n-able-n-central-zero-day-vulnerabilities
Happening right now?
Business email taken over, or a payment sent to the wrong account?
Call us now. The first hours matter most: your bank may still be able to recall the payment, and whoever is in the mailbox keeps access until it is locked down. We handle the lockout and help trace the payment.
Not an emergency? Get a free security assessment instead.
Protect Your Organization
Find vulnerabilities like this in your systems before attackers do.
24/7 monitoring to detect and respond to threats like these in real time.
Block phishing and malware delivery targeting your organization.
Map security controls to 26 frameworks including NIST, SOC 2, and HIPAA.