Return to Threats

Attackers Compromise Microsoft's Agent Framework PyPI Package to Steal Cloud Credentials

thehackernews.com 2026-11-18 AI supply chain Critical

Executive Summary

Attackers compromised Microsoft’s agent-framework package to distribute a credential stealer targeting cloud credentials. Teams running AI agents or Python-based AI applications should treat dependency provenance and version control as critical security requirements.

What Happened

Threat actors have compromised Microsoft's agent-framework package to push a credential stealer, according to reports from Aikido, Chainguard, Endor Labs, OX Security, SafeDep, Socket, StepSecurity, and Google-owned Wiz. The malicious versions, 1.26.1 and 1.26.2, were published 33 minutes apart on November 17, 2026, and are designed to run a credential stealer on every Python start. As of

Affected Products

  • Microsoft’s agent-framework package
  • PyPI

Vulnerability / Attack Technique

Malicious package compromise and credential stealing

Technical Mechanism

Malicious versions 1.26.1 and 1.26.2 were published 33 minutes apart on November 17, 2026. The versions are designed to run a credential stealer on every Python start.

AI Security Category

LLM03:2025 Supply Chain

Part of the AI Supply Chain and AI SBOM topic.

Relevant Frameworks

LLM03:2025 Supply ChainMITRE ATLAS: Credential AccessMITRE ATLAS: Execution

RealGround Analysis

This incident matters because AI agents commonly depend on third-party libraries that execute within developer or production environments. A compromised dependency can expose cloud credentials used by agents and LLM applications, so teams need continuous dependency visibility, provenance checks, and controlled update processes.

Recommended Controls

  • Maintain an inventory of AI-agent and LLM dependencies, including transitive Python packages.
  • Pin approved package versions and block versions not validated through security review.
  • Verify package provenance and integrity before installation or deployment.
  • Scan dependency updates for malicious behavior in isolated environments.
  • Use short-lived, least-privilege cloud credentials for AI-agent workloads and rotate credentials exposed to compromised packages.

Related RealGround Service

AI Agent SBOM and LLM Mapping · or start with the free AI Agent Security Scanner.

Sources

thehackernews.com: https://thehackernews.com/2026/11/attackers-compromise-microsofts-agent.html

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

Talk to AI CISO