CVE Index中文
L4 · Septic Tank BlewWindows · MeshAgent · GitHub · SharePointshadowcloud-windows-backdoors

ShadowCloud backdoors use GitHub, Ably and SharePoint on compromised networks

Kaspersky documented an active intrusion campaign primarily affecting government networks in Asia and Africa. Attackers used MeshAgent on compromised systems to deploy Windows backdoors that used GitHub, Ably and SharePoint for command and control.

Disclosed
Nov 17, 2026
Public disclosure date
Severity
Septic Tank Blew
No CVE: a breach, not a bug
Splash zone
Windows systems with unauthorized MeshAgent activity or correlated ShadowCloud loader and network indicators.
Windows · MeshAgent · GitHub · SharePoint
In the wild
Yes
Kaspersky observed multiple backdoor builds and victims in government-focused intrusions.

What happened

Kaspersky identified the campaign in early 2026. It found two instances of MeshAgent launched inside a Taiwanese organization. The agent connected to a configured external server and delivered a legitimate executable, a malicious DLL and an encrypted payload. The executable loaded the DLL, which unpacked the backdoor. The initial access method is unclear; the researchers found evidence consistent with a prior infection. The report does not establish a vulnerability in MeshAgent or compromise of the cloud providers.

GitDoor used GitHub repositories and Ably messaging to exchange data with the operator. GraphDoor used embedded application credentials to obtain a Microsoft Graph token and exchanged commands and results through SharePoint. Kaspersky recovered plug-ins for screen capture, keystroke collection and proxying. The observed victims were primarily government organizations across Asia and Africa. No total victim count was given.

Crime-scene timeline

  1. Nov 17, 08:00 UTC

    Kaspersky publishes the ShadowCloud investigation

    Two technical reports describe GitDoor, GraphDoor, the cloud command channels and supporting infrastructure.

Who said what

We list each claim and who made it. Primary sources (the vendor, the researcher, the registry) are marked. Where reports disagree, we grade by the worst credible case.