Attackers Compromise Microsoft's Agent Framework PyPI Package to Steal Cloud Credentials
Microsoft's official Python agent-framework package on PyPI was hijacked to distribute credential-stealing malware targeting Azure and CI pipelines.
Velocity
How fast coverage is spreading — measured hourly from article rate × source diversity. How this works →
The brief
Coverage from The Register, BleepingComputer, SecurityWeek, OX Security, wiz.io, and The Hacker News reports that attackers compromised Microsoft's official agent-framework Python package hosted on the Python Package Index (PyPI) to distribute malware designed to steal cloud credentials. The compromised package is Microsoft's official Python agent framework, a legitimate software component used by developers building AI agents and automated workflows. According to the reporting, the malicious code embedded in the poisoned releases specifically targets Azure credentials and continuous integration (CI) pipeline secrets, exfiltrating them to attacker-controlled infrastructure. The Register frames the incident as Microsoft's official Python agent framework being hijacked to distribute malware, while BleepingComputer emphasizes the credential theft aspect targeting cloud environments. SecurityWeek describes the releases as poisoned with a credential stealer, and OX Security's breaking report highlights the theft of both Azure and CI credentials. Wiz.io's analysis notes the malware targets Azure and CI pipelines specifically, and The Hacker News characterizes the attack as a compromise of the PyPI package to steal cloud credentials.
The coverage emphasizes the supply-chain nature of the attack, where a legitimate, Microsoft-maintained package on a major public repository was subverted to deliver malicious payloads to downstream users. Multiple security vendors — OX Security and Wiz — have published technical analyses, suggesting the compromise was detected through automated scanning or threat intelligence sharing. The Register, BleepingComputer, SecurityWeek, and The Hacker News provide journalistic context and amplification. The simultaneous publication timing across outlets suggests a coordinated vulnerability disclosure process, though the coverage does not specify which entity discovered the compromise first or whether Microsoft's security team was notified prior to public release. The malicious functionality is consistently described as credential theft aimed at Azure cloud resources and CI/CD pipeline secrets, which are high-value targets for lateral movement and further supply-chain attacks.
Background context includes the growing reliance on Python packages for AI agent development and the increasing frequency of supply-chain attacks targeting popular open-source repositories. PyPI has been a recurring vector for typosquatting, dependency confusion, and account takeover attacks, but the compromise of an official Microsoft-maintained package represents a significant escalation in attacker sophistication and impact radius. The agent-framework package is part of Microsoft's ecosystem for building autonomous AI agents, a rapidly expanding area of enterprise software development. Credential theft targeting Azure and CI pipelines aligns with known attacker objectives: gaining persistent access to cloud infrastructure, compromising build systems, and enabling further supply-chain contamination. The coverage does not specify the exact versions affected, the duration of the compromise, the number of downloads of poisoned releases, or whether any confirmed breaches have resulted from the malicious package.
What to watch next includes Microsoft's official security advisory and any PyPI takedown or quarantine actions for the affected package versions. Security teams should monitor for indicators of compromise related to the credential exfiltration infrastructure and audit any systems that installed the agent-framework package during the compromise window. The coverage does not state whether a patched version has been released or what remediation steps Microsoft recommends. Further technical analysis from OX Security, Wiz, or other vendors may reveal the initial access vector — whether the compromise stemmed from a compromised maintainer account, a compromised build pipeline, or another mechanism. Any downstream impact on organizations that consumed the poisoned package will likely emerge in subsequent incident reports.
Synthesized by PULSE from the headlines below under a strict no-invention contract. ✓ fact-checked: unsupported claims removed (95% supported) Updated 1h ago.
Quick answers
Which package was compromised?
Microsoft's official agent-framework Python package on PyPI.
What does the malware steal?
Azure credentials and CI/CD pipeline secrets.
When was this disclosed?
All six outlets published at 06:45 UTC on 18 November 2026.
Coverage (7)
- Two malicious agent-framework releases hit PyPI in 33 minutes Aikido Security · 14h ago
- Microsoft's official Python agent framework hijacked to distribute malware The Register · 14h ago
- Microsoft’s Agent Framework PyPI package compromised to steal credentials BleepingComputer · 14h ago
- Microsoft Agent Framework Releases Poisoned With Credential Stealer SecurityWeek · 14h ago
- BREAKING: Hijacked Releases of Microsoft’s Official agent-framework Steals Azure and CI Credentials OX Security · 14h ago
- Microsoft agent-framework Compromised: Credential Stealer Targets Azure and CI Pipelines wiz.io · 14h ago
- Attackers Compromise Microsoft's Agent Framework PyPI Package to Steal Cloud Credentials The Hacker News · 14h ago
Topics
Related trends
Firefox opens hotline for Windows users fed up with Microsoft shoving Edge down their throat
Mozilla launches a public hotline for Windows users frustrated by Microsoft's aggressive Edge promotion tactics.
US tech outages blamed on rush to fix bugs found by AI
Major U.S. tech firms and government agencies experienced simultaneous outages, prompting officials to deny cyberattacks while analysts question if rushed AI-driven bug fixes are the cause.
Social media algorithms are encouraging Americans to freak out about the tech outages
A string of tech outages and software glitches has sparked online panic, but experts say simple explanations apply.
How to Tell If It’s a Cyberattack or Just an Outage
A wave of tech outages across air travel, Microsoft services, and banking has left the public asking whether failures are accidents or attacks.
Google: ‘The era of autonomous attacks has arrived’
Google warned that the era of autonomous cyberattacks has arrived after discovering an AI framework exploiting software flaws without human intervention.
Hijacked Optimizely Scripts Served Card Skimmer and iOS Exploit Kits to CNN, BBC and Zoom Visitors
Hackers exploited an Optimizely vulnerability to serve malicious code on 178 major websites including CNN, BBC, Zoom, HSBC, Salesforce and Equifax.