AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: A Closer Look At The PyPI Hijack Of Microsoft’s AI Agent Framework on ThorstenMeyerAI.com

Before you orderOffer from Amazon

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
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

Two versions of Microsoft’s official Python package agent-framework, 1.26.1 and 1.26.2, were published on PyPI on November 17, 2026, with credential-stealing code. PyPI quarantined both releases; the latest clean version identified in the advisories is 1.26.0. Anyone who installed either affected release should treat the environment as exposed, rebuild it from a clean base and rotate credentials it could access.

Two releases of Microsoft’s official Python package for its Agent Framework contained credential-stealing code and were briefly available on PyPI on November 17, according to security advisories and researchers. The affected versions, 1.26.1 and 1.26.2, could run malicious code when Python started in an environment where either release had been installed, putting accessible cloud, developer and CI credentials at risk.

The GitHub Advisory Database says the malicious code was absent from the preceding clean release, 1.26.0, and rates the incident critical. PyPI records cited in the source report place version 1.26.1’s publication at 20:35 UTC and version 1.26.2’s at 21:08 UTC on November 17. StepSecurity says PyPI quarantined the first at 21:06 UTC and the second at 21:16 UTC. The report describes the first release as available for about 31 minutes; the second was quarantined eight minutes after publication.

The package’s normal framework functionality continued to work, according to the advisory, so successful installation or ordinary use would not necessarily reveal the compromise. The malicious code added a Python start-up hook that could run when the interpreter started, without the user importing the Agent Framework. The advisory describes collection of credentials and tokens accessible to the affected environment, including items associated with cloud services, code hosting, AI services and package registries.

Researchers also reported differences in the second release. StepSecurity says it added capabilities, including requests for GitHub Actions identity tokens. The advisory and researchers’ analyses are the basis for these technical findings; they do not establish how many systems installed the packages or how many credentials were successfully taken. PyPI removed both versions, and the source report identifies 1.26.0, released November 12, as the current clean version.

At a glance
reportWhen: Published November 17, 2026; both relea…
The developmentSecurity researchers and advisories reported that two compromised releases of Microsoft’s Agent Framework package on PyPI collected credentials before the registry quarantined them.

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

  1. Clean 1.26.0 publishedStill the current version on PyPI
  2. Malicious 1.26.1 appearsFramework works normally, a credential stealer runs whenever Python starts
  3. PyPI quarantines 1.26.1After about 31 minutes
  4. Attacker uploads 1.26.2Renamed files, also targets GitHub Actions tokens
  5. PyPI quarantines 1.26.2Eight minutes later
  6. Malware databases list itMAL-2026-20101 (OpenSSF), GHSA-764m-3ppc-5m73 (GitHub)
Quarantine is not revocation. The second release came 2 minutes 23 seconds after the first was pulled: the attacker still had publishing access. Copies already in caches, mirrors and images stay there.

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

Credentials at Risk in Build Environments

The incident matters because Python packages are often installed automatically in developer workstations, build servers, CI pipelines and container images. A malicious package in any of those places may gain access to secrets used to publish software or connect to cloud services. The advisory specifically calls for rotating credentials the affected environment could reach, including CI and GitHub Actions secrets.

The code’s reported start-up behavior widened the risk beyond a developer deliberately running the framework. Any later Python process in an affected environment could trigger it, even if the agent package itself was not imported. The framework continuing to function also meant users might not notice a problem through ordinary application behavior.

There is a separate registry-security concern: StepSecurity says quarantining the first release did not immediately revoke the publisher’s access, allowing a second release to appear shortly afterward. JFrog similarly said quarantining a version does not itself invalidate a compromised publishing credential. That sequence highlights why incident response may need to address both malicious artifacts and the account or credential used to publish them.

Products Worth Considering

DJI Mini 4K Drone Combo with 4K Camera, 3-Axis Gimbal, and Two Batteries

DJI Mini 4K Drone Combo with 4K Camera, 3-Axis Gimbal, and Two Batteries

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 workflows in Python. The PyPI package agent-framework is an umbrella package that brings in the framework’s core code and integrations. The source report says Microsoft’s own installation instructions direct users to this package, distinguishing it from an unrelated lookalike.

Microsoft published version 1.26.0 on November 12. Five days later, version 1.26.1 appeared on PyPI as a wheel-only release, according to the GitHub advisory. After PyPI quarantined it, an attacker uploaded version 1.26.2 roughly two minutes later, according to StepSecurity; PyPI quarantined that version at 21:16 UTC. The OpenSSF malware database listed the case as MAL-2026-20101, and the GitHub Advisory Database published GHSA-764m-3ppc-5m73.

Security firms including Socket, StepSecurity, Wiz, JFrog and Endor Labs examined the releases. Wiz said it promptly reported its findings to Microsoft on November 17. Microsoft told The Register: “We have made some repositories private while we investigate malicious content.” The source report said it had found no Microsoft security advisory at publication time.

“The credential stealer is not present in the last clean release, 1.26.0.”

— GitHub Advisory Database

Victim Count and Exposure Remain Unknown

The source material does not give a confirmed number of downloads, affected installations, compromised accounts or successfully exfiltrated credentials. Wiz says the package receives almost 200,000 downloads a month, but that figure is not a count of downloads of the malicious releases, and the comparison basis for exposure during the short publication window is not provided.

It is also not clear from the available information how the publishing access was compromised, who was responsible, or whether stolen credentials were used after collection. Researchers found no remote-code-loading second stage in the inspected code, and The Register reported that the malware was not persistent beyond the installed package; those findings do not determine whether credentials were taken or misused. Microsoft has not provided a detailed account in the cited material, and the precise scope of its investigation remains unknown.

Rebuild and Rotate Accessible Secrets

Organizations should check package and build records for installations of agent-framework 1.26.1 or 1.26.2. The source report relays the advisory’s recommendation to treat any environment that installed either version as exposed, rebuild it from a clean base using version 1.26.0, and rotate every credential that environment could access. That includes secrets in CI systems and GitHub Actions, as well as cloud and service credentials available to the machine or pipeline.

Teams should also review their own logs and security procedures for evidence of access or use of credentials, following the detailed guidance in the official advisories rather than relying on the package’s continued normal operation as reassurance. The source report says the advisories contain technical details for defenders, but this article does not reproduce indicators or attack instructions.

Further clarity depends on Microsoft’s investigation and any additional advisories from Microsoft, PyPI or security researchers. Those updates may establish how the publishing access was obtained, how many environments installed the releases and whether any exposed credentials were used. Until then, affected users should act on installation history and access to secrets, rather than waiting for a confirmed victim count.

Key Questions

Which package versions were affected?

The affected PyPI releases were agent-framework 1.26.1 and 1.26.2. The source report identifies 1.26.0, released November 12, as the preceding clean version and the current version on PyPI at publication.

How did the malicious code run?

The GitHub advisory says the releases added a Python start-up hook that could run when Python started in an affected environment, without importing agent_framework. The framework itself could continue working normally.

What should an organization do if it installed either release?

Treat the environment as exposed, rebuild it from a clean base on version 1.26.0, and rotate every credential the environment could access, including CI and GitHub Actions secrets, following the relevant security advisories.

How many users or systems were affected?

No confirmed victim or installation count is given in the source material. Wiz’s estimate of almost 200,000 monthly downloads refers to the package generally, not to downloads of the malicious versions.

Has Microsoft explained how the package was compromised?

Not in the cited material. Microsoft told The Register it had made some repositories private while investigating malicious content, but the source report said no detailed Microsoft security advisory had been found at publication time.

Source: ThorstenMeyerAI.com

You May Also Like

What It Takes To Trust AI Agents: Insights From Mythos 5

UK and Anthropic reports detail how Claude Mythos 5 targeted real developers during a cyber test and why it failed to stop.

How Advanced Particle Geometry Mapping Fuels AI Breakthroughs In ‘SINGULARITY’

Innovative particle geometry techniques are accelerating AI development in ‘SINGULARITY,’ enabling more immersive and intelligent environments.

Explanation Of Everything You Can See In Htop/top On Linux (2019)

Detailed explanation of all elements displayed in htop and top commands on Linux, clarifying what each component means and how to interpret system performance.

2026’S AI Landscape: 9 Key Innovations To Know

An overview of the nine most significant AI innovations expected in 2026, highlighting confirmed developments and their impact on technology and society.