🔍 Read the full analysis: How Developers Can Respond To The Microsoft AI Agent Framework PyPI Hijack 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 releases of Microsoft’s official Python package agent-framework, versions 1.26.1 and 1.26.2, contained credential-stealing code and were quarantined by PyPI on November 17, 2026. Security advisories say the code could run when Python started, even without the framework being imported. Developers who installed either version should rebuild affected environments and rotate credentials they could access.
Two malicious versions of Microsoft’s official agent-framework Python package were briefly available on PyPI on November 17, and security advisories warn that they contained code designed to steal credentials. Developers and organizations whose machines installed versions 1.26.1 or 1.26.2 should treat those environments as exposed, rebuild them from a clean base, and rotate credentials the environments could reach.
The GitHub Advisory Database says versions 1.26.1 and 1.26.2 contained a credential stealer absent from the last clean release, 1.26.0. It rates the incident critical. The package is the one Microsoft’s installation instructions point to, rather than an unrelated copycat. The advisory says the malicious code could run when Python started in an affected environment, without a user importing the agent framework itself.
According to the timelines reported by StepSecurity, version 1.26.1 was published at 20:35 UTC and quarantined by PyPI at 21:06 UTC. An attacker uploaded version 1.26.2 at 21:08 UTC, roughly two minutes after the first release was quarantined; PyPI quarantined that version at 21:16 UTC. Both releases have been removed from the registry. The current PyPI version is reported as 1.26.0, released November 12.
The advisory describes collection of credentials and tokens from affected systems, including secrets associated with cloud services, developer tools, and build pipelines. It says version 1.26.2 added the ability to request GitHub Actions identity tokens. The package’s ordinary framework functions continued to work, which could make the compromise difficult to notice from routine use. Security firms including Socket, StepSecurity, Wiz, JFrog and Endor Labs analyzed the releases.
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
Risks to Developer Credentials
The incident matters because Python packages can run in developer laptops, automated build systems, CI pipelines and container images. A compromise in a package used by those systems can put more than a single user account at risk: cloud credentials and pipeline tokens may grant access to source code, deployment systems or hosted services. The advisory’s warning applies even when the framework was not explicitly imported, because the reported code ran at Python startup.
PyPI’s quarantine stopped the flagged releases from being installed from the registry, but it does not remove copies already installed or invalidate credentials that may have been collected. StepSecurity said quarantining the first version did not immediately revoke the attacker’s publishing access, and the later upload followed before the second quarantine. That sequence highlights why incident response must address both affected environments and credentials, not just package availability.
The recommended response is to identify systems that installed either release, rebuild them from a clean base using the last clean version, and rotate every credential the affected environment could access. The GitHub advisory specifically includes CI and GitHub Actions secrets in its warning. Organizations should also review their own logs and response procedures; the public advisories are not a substitute for determining what a particular system could access.
Products Worth Considering

Bose QuietComfort Wireless Bluetooth Noise-Cancelling Headphones, Cypress Green

DJI Mini 4K Camera Drone Combo with 2 Batteries

Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation
As an affiliate, we earn on qualifying purchases.
How the Releases Reached PyPI
Microsoft Agent Framework is an open-source Python toolkit for building AI agents and coordinating them in workflows. The package named agent-framework is an umbrella package that pulls in the framework’s core code and integrations. According to the GitHub advisory, the malicious releases differed from the clean 1.26.0 release published on November 12.
StepSecurity reported that PyPI quarantined 1.26.1 at 21:06 UTC on November 17, then that 1.26.2 appeared at 21:08 UTC and was quarantined at 21:16 UTC. Researchers described the first release as a wheel-only upload; Endor Labs said releases it examined from 1.21.0 through 1.26.0 had also included source code. The OpenSSF malware database listed the case as MAL-2026-20101; the GitHub Advisory Database published advisory GHSA-764m-3ppc-5m73 on November 18.
Microsoft’s response reported to The Register was limited to a short statement: “We have made some repositories private while we investigate malicious content.” The source report said it had not found a Microsoft security advisory at publication time. Wiz said it promptly reported its findings to Microsoft on November 17.
““We have made some repositories private while we investigate malicious content.””
— Microsoft, in a statement to The Register
Victim Scope Remains Unknown
The available source material does not establish how many systems installed the two releases, whether credentials were successfully used after collection, or which organizations were affected. A package download count is not the same as a count of installations or compromised environments. Wiz was reported as saying the package receives almost 200,000 downloads a month, but that figure does not identify how many downloads occurred during the incident window.
It also remains unclear from the cited material how the attacker obtained publishing access, whether any stolen credentials were used, and what Microsoft’s investigation has found. StepSecurity and JFrog both emphasized that quarantining a release does not, by itself, revoke compromised publishing credentials. The source report also notes that StepSecurity found no remote-code-loading second stage in the code it inspected and that the malware was not persistent beyond the installed package; those findings do not establish whether data had already been exfiltrated.
Rebuild, Rotate, and Monitor
Developers and security teams should first check dependency records, lockfiles, build logs and deployment artifacts for agent-framework 1.26.1 or 1.26.2. Any environment that installed either release should be treated as exposed, according to the advisory. The recommended response is to rebuild from a clean base using version 1.26.0 and rotate all credentials that the affected system could access, including cloud, source-control, package-registry and CI credentials.
Teams should review relevant authentication and pipeline logs for activity that they cannot account for, following their established incident-response procedures. They should also check whether affected package versions entered cached dependencies or container images, rather than checking only current installations. These actions are defensive steps based on the advisory’s scope; the source material does not confirm that every affected environment was compromised or that every credential was stolen.
Further updates depend on the findings of Microsoft and the security researchers, including any additional advisories, details on the publishing-account compromise, and information about victim impact. Until that information is available, the published guidance is to remove affected environments from use, rebuild them cleanly and rotate reachable secrets.
Key Questions
Which package versions were affected?
The GitHub Advisory Database identifies agent-framework 1.26.1 and 1.26.2 as malicious. It names 1.26.0, released November 12, as the last clean release in the cited timeline.
Could the malicious code run if the framework was never imported?
Yes, according to the advisory. It says the code ran when Python started in an affected environment, without requiring an import of agent_framework.
What should a team do if it installed one of the releases?
Treat the environment as exposed, rebuild it from a clean base on the clean release, and rotate every credential it could access. The advisory specifically calls out CI and GitHub Actions secrets.
How many organizations were affected?
The cited material does not provide a confirmed victim count. The reported monthly download figure is not a count of installations during the incident or of compromised systems.
Does PyPI’s quarantine remove the risk?
Quarantine prevents installation of the flagged releases from PyPI, but it does not clean copies already installed or invalidate credentials that may have been exposed. Affected systems and accessible credentials still require investigation and remediation.
Source: ThorstenMeyerAI.com
