Skip to content
NewA calmer, faster ZeroHour

Poisoned agent-framework PyPI releases steal cloud and CI credentials

Original title: Microsoft Agent Framework Releases Poisoned With Credential Stealer

SecurityWeekby Eduard Kovacs ()2 min read
Part of a story covered by 9 sourcesPyPI quarantines malicious Microsoft agent-framework 1.26.1 and 1.26.2Merged summary & timeline
AI summary · x-ai/grok-4.7Summaries by Vani AI

Versions 1.26.1 and 1.26.2 were quarantined; 1.26.0 remains the latest clean release of a package with about 190,000 monthly downloads.

Two malicious versions of Microsoft’s Agent Framework Python package, agent-framework, were published on PyPI on November 17 about 33 minutes apart. The builds kept legitimate components and added code that runs when the Python interpreter starts, even without an explicit import, and tries to steal environment, file, process, Azure, and later GitHub Actions credentials. Both versions were quarantined. Microsoft said an investigation was underway. How the packages were published is unknown.

Unconfirmed: How the packages were published, whether the vibiret account uploaded them, and whether other missing Microsoft repositories are related remain unconfirmed.

Key facts
  • Malicious versions were 1.26.1 and 1.26.2; 1.26.0 is the latest clean release.
  • The package has about 190,000 downloads per month.
  • Targets include Azure managed identity and Azure DevOps tokens, plus GitHub Actions secrets in the second release.
  • A PyPI account named vibiret had publishing rights, but its role is unconfirmed.
  • Microsoft’s Agent Framework GitHub repository was inaccessible when the article was written.
Security assessment · AI, from the article

Impact: Theft of cloud, CI/CD, and developer credentials

ProductAffectedFixed in
Microsoft agent-framework1.26.1 and 1.26.21.26.0
What to do
  1. 1.Check developer environments, CI/CD pipelines, and container images for versions 1.26.1 and 1.26.2.
  2. 2.Replace affected installations with verified packages and rebuild affected environments.
  3. 3.Rotate potentially exposed credentials and investigate suspicious cloud activity.
Analyze this story with Vani AIImpact, who is affected and what defenders should do, in a new chat

From the original article

2 min read at securityweek.com

Versions 1. 26. 1 and 1.

SecurityWeek

By Eduard Kovacs · securityweek.com

Read the full article at SecurityWeek

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.