Hermes Vector

Tech

Microsoft Python agent framework hijacked to distribute malware

Malicious versions 1.26.1 and 1.26.2 of Microsoft's agent-framework package on PyPI steal developer secrets and request fresh cloud identity tokens.

By Hermes Vector

Also in: Español · Português · Français · Deutsch · हिन्दी

Checked against 3 sourcestheregister.comnews.lavx.huphoneworld.com.pk
Illustration for: Microsoft Python agent framework hijacked to distribute malware

In brief

  • Malicious versions 1.26.1 and 1.26.2 of Microsoft's agent-framework package were uploaded to PyPI.
  • The malware steals environment variables, credential files, and secrets for Azure, Git, Docker, and Hugging Face.
  • It also requests fresh identity tokens from GitHub Actions, Azure DevOps, and Azure.
  • Microsoft made some GitHub repositories private while investigating malicious content.

Microsoft Python agent framework hijacked to distribute malware

Microsoft's official Python agent-framework package on PyPI has been hijacked to distribute information-stealing malware, according to multiple security research firms. The affected releases are versions 1.26.1 and 1.26.2.

What happened

Security researchers said a compromised administrator account on Microsoft's GitHub organization allowed malicious versions of the package to be uploaded to the Python Package Index. OX Security reported that version 1.26.1 was reverted to a safe release, but the attacker later uploaded version 1.26.2 through Microsoft's official PyPI publisher account, suggesting the credentials had not been locked down at the time of the report.

The discovery emerged on Wednesday, Nov. 18, 2026, during Microsoft's Ignite conference, where CEO Satya Nadella appeared on stage.

What the malware does

The malicious package does not need to be imported into a project to activate. Simply installing it and starting a Python interpreter is enough to run the harmful code. The package pulls the genuine 1.26.0 release of Microsoft's agent-framework so it continues to work and appears legitimate while it collects secrets.

The malware gathers environment variables, credential files, and other secrets for Azure, Git, Docker, Hugging Face, and other developer services. It encodes the data as a base64-encoded chunk of JSON and sends it to a hard-coded webhook.site address.

The malware also requests fresh identity tokens from GitHub Actions, Azure DevOps, and Azure, then sends those tokens to the same webhook.site address. Researchers at OpenSourceMalware said that combination of passive theft and active token requests makes it especially dangerous in cloud build environments, because even short-lived tokens are useful when they are exfiltrated right away.

The malware is not persistent and runs only while the malicious release remains in a site-packages directory.

Microsoft's response

The agent-framework GitHub repository is offline. Microsoft also took the ONNX Runtime repository offline, but neither Microsoft nor security researchers have linked that repository's removal to the malware campaign.

"We have made some repositories private while we investigate malicious content," a Microsoft spokesperson told The Register. "Our priority is to protect customers and the broader ecosystem. We are working to review and will take further action as needed."

Why it matters

Microsoft agent-framework for Python draws close to 200,000 downloads a month and is used to build and deploy AI agents. Teams that installed versions 1.26.1 or 1.26.2 should treat their environment as compromised, remove the malicious release, reinstall a known-safe release, and rotate secrets for Azure, Git, Docker, Hugging Face, and other services.


Sources

This article was drafted with AI assistance and checked against the sources above. Company claims are reported as claims. Cover image is AI-generated.

Questions readers ask

What happened to Microsoft's Python agent-framework package?
Microsoft's official Python agent-framework package on PyPI was hijacked to distribute information-stealing malware in versions 1.26.1 and 1.26.2.
When was the compromise discovered?
The discovery emerged on Wednesday, Nov. 18, 2026, during Microsoft's Ignite conference.
How does the malware operate?
The malware activates simply by installing the package and starting a Python interpreter, then collects secrets and tokens to send to a hard-coded webhook.site address.
Why is this incident significant for users?
The package draws close to 200,000 downloads a month, so teams that installed versions 1.26.1 or 1.26.2 should treat their environment as compromised and rotate secrets.

The day's checked news, by email

One short email a day with the top stories. Free, and you can unsubscribe in one click.

Or follow for the latest news:

How do you feel about this story?

Comments 0

No comments yet. Start the discussion.

Tech

Intel signs first external customer for 18A chips

Intel CEO Lip-Bu Tan announced the company signed its first outside customer for its 18A manufacturing process, a milestone in its five-year effort to compete with TSMC in contract chipmaking.

Checked against 6 sourcesCNBCtechbuzz.aifinance.biggo.com+3 more

Tech

Oregon Sues OpenAI for $2.035 Billion Over ChatGPT Child Safety

Oregon Attorney General Dan Rayfield sued OpenAI on Nov. 17, 2026, alleging ChatGPT endangers children and the company misled the public about its safety. The state seeks at least $2.035 billion in damages and civil penalties.

Checked against 4 sourcesengadget.comqz.comtechrepublic.com+1 more