Return to Threats

New GitDoor and GraphDoor Backdoors Turn GitHub and SharePoint Into Covert C2 Channels

thehackernews.com 2026-11-17 AI agent abuse High

Executive Summary

Operation ShadowCloud reportedly targeted government sectors in Asia and Africa using GitDoor and GraphDoor backdoors that leveraged GitHub and Microsoft SharePoint for command and control. Organizations operating AI agents or LLM applications should care because these services may be trusted integrations or data sources.

What Happened

Government sectors in Asia and Africa have been targeted by an active cyber intrusion campaign that delivers to previously undocumented backdoors that leverage legitimate services like Microsoft SharePoint and GitHub for command‑and‑control (C2). The activity has been codenamed Operation ShadowCloud by Kaspersky. The Russian cybersecurity company said it identified the campaign in early 2026.

Affected Products

  • Microsoft SharePoint
  • GitHub
  • GitDoor
  • GraphDoor

Vulnerability / Attack Technique

Covert command-and-control through legitimate services

Technical Mechanism

The reported backdoors leverage legitimate Microsoft SharePoint and GitHub services for command-and-control communications. The summary does not provide further technical details about how GitDoor or GraphDoor authenticate, exchange commands, or execute payloads.

AI Security Category

AI-assisted attack

Part of the AI-Assisted Attacks topic.

RealGround Analysis

This activity is not described as exploiting an AI or LLM weakness, but it is relevant to AI-agent environments that connect to GitHub or SharePoint. Agents with excessive permissions or insufficient validation could potentially amplify the impact of compromised service content; this implication is RealGround analysis, not a reported fact.

Recommended Controls

  • Restrict AI-agent access to GitHub and SharePoint to explicitly approved repositories, sites, and operations.
  • Require least-privilege, short-lived credentials and separate service identities for AI-agent integrations.
  • Monitor agent-originated requests and service activity for anomalous commands, repositories, files, or access patterns.
  • Validate and sanitize external content before allowing it to influence agent actions or tool calls.
  • Include GitHub and SharePoint integrations in AI red-team exercises and business-logic audits.

Related RealGround Service

AI Agent Business Logic Audit · or start with the free AI Agent Security Scanner.

Sources

thehackernews.com: https://thehackernews.com/2026/11/new-gitdoor-and-graphdoor-backdoors.html

Summary and analysis written by RealGround from the source above; last updated 2026-11-18.

Talk to AI CISO