Ethical Hacking News
Microsoft's Agent Framework PyPI package has been compromised, leaving cloud users and AI developers at risk of data breaches and credential theft. The malicious package, version 1.26.1 and 1.26.2, was published on November 17, 2026, and is designed to steal cloud credentials and AI-related data. To mitigate the threat, users are advised to take immediate action to isolate affected hosts and runners, pin the package to the most recent safe version, and check for exposure to the malicious versions.
A recent vulnerability in Microsoft's Agent Framework PyPI package has left cloud users on high alert. The compromised package, version 1.26.1 and 1.26.2, was published on November 17, 2026, and is designed to steal cloud credentials and AI-related data. The malicious package installs Python startup hooks that silently collect credentials and exfiltrate them to a webhook endpoint. The package also drops a .pth file in the site-packages root directory, causing the credential stealer to run automatically on every interpreter start. The compromised package attracted about 190,000 downloads a month, making it a significant threat to cloud users. Users are advised to perform specific steps to mitigate the vulnerability, including isolating affected hosts and runners. The attack highlights the growing threat of supply chain attacks and the need for cloud users to remain vigilant in protecting their credentials and data.
A recent vulnerability in Microsoft's Agent Framework PyPI package has left security experts and cloud users on high alert. The compromised package, version 1.26.1 and 1.26.2, was published on November 17, 2026, and is designed to steal cloud credentials and AI-related data. The malicious package was created by threat actors who managed to obtain publishing access, allowing them to push a credential stealer onto the package.
The compromised package, agent-framework, is a meta-package for Microsoft Agent Framework, an open-source framework for building and orchestrating artificial intelligence (AI) agents using Python. The malicious versions of the package install Python startup hooks that silently collect credentials, environment data, and cloud identity tokens, then exfiltrate them to a webhook endpoint. The package also drops a .pth file in the site-packages root directory, which then gets executed automatically on every interpreter start, causing the credential stealer to run whenever Python is started in an environment, even if the package is not explicitly imported.
The malicious package was able to capture a wide range of data, including environment variables based on substring-based matching, credential files associated with Azure CLI, and Azure DevOps OIDC tokens. The package also asked the platform to issue new credentials, which would not be covered by rotating stored secrets. The collected information was saved into a JSON document, zlib-compressed, Base64-encoded, and sent via an HTTPS POST request to a hard-coded URL.
The compromised package attracted about 190,000 downloads a month, making it a significant threat to cloud users. To mitigate this vulnerability, users are advised to perform the following steps: isolate affected hosts and runners, pin the package to the most recent safe version (1.26.0) and restore from a clean environment, check for exposure to the malicious versions between 8:35 and roughly 9:20 p.m. UTC on November 17, 2026, search proxy and DNS logs for requests to webhook[.]site/d381fd80-ec62-42ad-8f5c-d1b15103c9bb, rotate exposed credentials from a clean system, audit Azure identities and review sign-in and activity logs, and audit GitHub Actions workflows that deploy to Azure.
The compromised package is the second Microsoft-related PyPI package to be compromised this year, following TeamPCP, which pushed three malicious versions of durabletask, Microsoft's Python client for the Durable Task workflow execution framework, to distribute the Mini Shai-Hulud worm. The attack on the agent-framework package highlights the growing threat of supply chain attacks and the need for cloud users to remain vigilant in protecting their credentials and data.
The threat actors behind the attack have demonstrated a clear understanding of the vulnerability and have managed to exploit the publishing access to create a malicious package. The package's design and functionality demonstrate a clear intent to steal cloud credentials and AI-related data, highlighting the need for users to take immediate action to mitigate the threat.
In conclusion, the compromise of Microsoft's Agent Framework PyPI package is a significant threat to cloud users and AI developers. The malicious package's ability to steal cloud credentials and AI-related data makes it a serious concern, and users are advised to take immediate action to mitigate the threat. By following the recommended steps and staying vigilant, cloud users can minimize the risk of data breaches and protect their credentials and data.
Related Information:
https://www.ethicalhackingnews.com/articles/Microsofts-Agent-Framework-PyPI-Package-Compromised-A-Threat-to-Cloud-Credentials-and-AI-Developer-Ecosystem-ehn.shtml
https://thehackernews.com/2026/11/attackers-compromise-microsofts-agent.html
https://cybersecuritynews.com/microsofts-agent-framework-pypi-package/
Published: Wed Nov 18 02:33:23 2026 by llama3.2 3B Q4_K_M