🔍 Read the full analysis: Microsoft’s AI Agent Framework PyPI Hijack: A Security Response Question on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Two versions of Microsoft’s Python package agent-framework, 1.26.1 and 1.26.2, were briefly available on PyPI on November 17, 2026, with credential-stealing code, according to security advisories. PyPI quarantined both releases; anyone who installed them should treat the environment as exposed, rebuild from a clean version and rotate credentials it could access.
Two versions of Microsoft’s Agent Framework Python package—1.26.1 and 1.26.2—were briefly available on PyPI on November 17 with credential-stealing code, according to the GitHub Advisory Database and security researchers. PyPI quarantined both releases, but developers who installed either version should treat the affected environment as exposed and rotate credentials it could access.
The GitHub Advisory Database says the malicious releases added a credential stealer that was absent from the previous clean version, 1.26.0. It rates the incident “critical.” The package is Microsoft’s published agent-framework umbrella package, and Microsoft’s installation instructions point users to it. Security firm Wiz says it receives almost 200,000 downloads a month; that figure describes the package’s reported monthly downloads, not the number of affected installations.
According to timelines published by researchers, version 1.26.1 appeared at 20:35 UTC on November 17 and was quarantined by PyPI at 21:06 UTC. A second version, 1.26.2, was uploaded at 21:08 UTC and quarantined at 21:16 UTC, according to StepSecurity. The current PyPI release identified in the source material is 1.26.0, published November 12. Both malicious versions have been removed from the registry.
The advisory says the code ran when Python started in an affected environment, even if the user did not import or directly use the framework. It searched for credentials and tokens in developer and cloud environments, and the second release added further capabilities. Researchers say collected information was sent to an attacker-controlled address. The package’s ordinary framework functions continued to work, which could make the compromise less apparent to users.
agent-framework on PyPI · UTC
Two poisoned releases, one evening
The official package of Microsoft’s Agent Framework, step by step. Blue: what happened. Green: the clean version.
November 12 to 18, 2026
- Clean 1.26.0 publishedStill the current version on PyPI
- Malicious 1.26.1 appearsFramework works normally, a credential stealer runs whenever Python starts
- PyPI quarantines 1.26.1After about 31 minutes
- Attacker uploads 1.26.2Renamed files, also targets GitHub Actions tokens
- PyPI quarantines 1.26.2Eight minutes later
- Malware databases list itMAL-2026-20101 (OpenSSF), GHSA-764m-3ppc-5m73 (GitHub)
Sources: GitHub Advisory Database GHSA-764m-3ppc-5m73, Nov 18, 2026; OSV MAL-2026-20101, Nov 17, 2026; PyPI project data, checked Nov 19, 2026; StepSecurity analysis (quarantine times).
Graphic: Powered by Thorsten Meyer AI
Why Build Environments Face Risk
The exposure matters because Python packages are often installed automatically during development, testing and deployment. A compromised dependency in a developer laptop, CI pipeline or build server may have access to secrets that are more consequential than the package itself, including cloud credentials, source-control tokens and AI-service keys. The advisory specifically calls for rotating credentials the affected environment could reach, including CI and GitHub Actions secrets.
Researchers also highlighted the gap between quarantining a malicious release and stopping further uploads. StepSecurity reported that the second release appeared about two minutes after the first was quarantined. It said the quarantine did not immediately revoke the attacker’s publishing access. JFrog similarly noted that removing one version does not, by itself, invalidate a compromised publishing credential. The sequence makes account and publishing-access investigation a key part of the response, beyond removing the package from PyPI.
The incident does not establish that every user of the framework was affected. The immediate concern is narrower: environments that installed either malicious version during its availability may have exposed credentials. Since the package could run its hook on later Python starts, simply avoiding the framework’s features would not address an already affected environment.
Popular right now

KitchenAid Artisan 5-Quart Tilt-Head Stand Mixer, Almond Cream

Sony WH-1000XM6 Wireless Noise-Canceling Headphones, Black

Dyson Cyclone V10 Origin Cordless Vacuum
As an affiliate, we earn on qualifying purchases.
How the Two Releases Appeared
Microsoft’s Agent Framework is an open-source toolkit for building AI agents and coordinating them in workflows. The PyPI package agent-framework acts as an umbrella package that installs the core framework and integrations. The previous release, 1.26.0, was published on November 12. Endor Labs said releases it examined from 1.21.0 through 1.26.0 had included source code, while 1.26.1 was published as a wheel only.
Several security firms, including Socket, StepSecurity, Wiz, JFrog and Endor Labs, examined or reported the incident. The OpenSSF’s open-source malware database listed it as MAL-2026-20101 on November 17, and the GitHub Advisory Database published GHSA-764m-3ppc-5m73 early on November 18. These advisories provide technical information for defenders; this report does not reproduce indicators of compromise or attack instructions.
Microsoft’s response in the supplied reporting was limited to a statement to The Register: “We have made some repositories private while we investigate malicious content.” The source material said it had not found a Microsoft security advisory at the time of publication. That statement concerns repositories being made private; it does not by itself explain the cause of the PyPI publication or confirm the status of any credentials.
“The credential stealer is not present in the last clean release, 1.26.0.”
— GitHub Advisory Database
Victim Counts and Access Remain Unknown
The source material gives no confirmed count of affected installations, organizations or stolen credentials. The reported monthly download figure from Wiz is not a count of downloads during the exposure window and cannot establish how many users installed either malicious release. It is also not clear whether the attacker successfully used any collected credentials or accessed systems beyond the package publication itself.
The available reporting does not establish how the attacker obtained publishing access, whether any credentials were compromised, or the full scope of Microsoft’s investigation. Researchers reported no remote-code-loading second stage in the code they inspected, and The Register summarized their findings as indicating the malware was not persistent beyond the installed package. Those observations are bounded by the code examined and do not rule out other activity. The precise number of affected environments and the status of all potentially exposed secrets remain unknown.
Rebuild and Rotate Exposed Secrets
Organizations should identify whether any laptop, container image, build server or CI job installed agent-framework 1.26.1 or 1.26.2. The advisory recommends treating those environments as exposed, rebuilding from a clean base using version 1.26.0, and rotating every credential the environment could access. That includes cloud keys, package-registry credentials and CI or GitHub Actions secrets where applicable. Teams should follow the advisories for technical detection and response details.
Further updates depend on Microsoft’s investigation and any additional findings from PyPI or security researchers. Readers should watch for a fuller account of the publishing-access compromise, a confirmed assessment of affected users and guidance on whether any credentials were used after theft. Until those details are available, the confirmed release window and affected version numbers provide the clearest basis for checking exposure.
Key Questions
Which package versions were affected?
The GitHub Advisory Database identifies agent-framework 1.26.1 and 1.26.2 as containing credential-stealing code. Version 1.26.0 is identified as the last clean release in the supplied advisory material.
When were the malicious releases available?
Researchers say version 1.26.1 was published at 20:35 UTC on November 17, 2026, and quarantined at 21:06 UTC. Version 1.26.2 appeared at 21:08 UTC and was quarantined at 21:16 UTC. Both releases have been removed from PyPI.
What should developers do if they installed one of the versions?
The advisory recommends treating the environment as exposed, rebuilding it from a clean base on version 1.26.0, and rotating all credentials it could access, including relevant CI and GitHub Actions secrets.
How many users were affected?
No confirmed victim or installation count is provided in the source material. Wiz’s estimate of almost 200,000 downloads a month is a general monthly figure, not a count of installs of the malicious releases.
Has Microsoft explained how the package was compromised?
Not in the statement cited in the source material. Microsoft told The Register it had made some repositories private while investigating malicious content. The cause of the publishing-access compromise and the investigation’s full findings remain unclear.
Source: ThorstenMeyerAI.com
