🔍 Read the full analysis: Microsoft’s AI Agent Framework Was Hijacked On PyPI To Steal Credentials. Who Has To Act 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 official Python package agent-framework, 1.26.1 and 1.26.2, contained credential-stealing code and were published on PyPI on November 17, 2026. PyPI quarantined both releases, but anyone who installed either should treat the environment as exposed, rebuild from a clean base, and rotate credentials it could access. The number of affected installations and whether stolen credentials were used remain unknown.
Two releases of Microsoft’s official Python package for its Agent Framework contained code that sought developer and cloud credentials before PyPI quarantined them on November 17, according to security advisories and researchers. Versions 1.26.1 and 1.26.2 were available for a short period, and the framework continued to work normally, making the malicious behavior less likely to be obvious to users.
The GitHub Advisory Database says the malicious code was absent from the preceding clean release, version 1.26.0. PyPI records cited by the source article show version 1.26.1 was uploaded at 20:35 UTC. StepSecurity says PyPI quarantined it at 21:06 UTC. A second release, version 1.26.2, appeared at 21:08 UTC and was quarantined at 21:16 UTC, according to the same security firm’s timeline. Both releases have since been removed from the registry.
The advisory says a Python start-up hook triggered the code when Python launched in an affected environment, even if the user did not import the Agent Framework. The package itself continued functioning. Researchers said the code sought credentials in environment variables and common developer-tool files, and could request cloud or pipeline identity tokens. The advisory describes collection of credentials associated with services including Azure, AWS, GitHub, Hugging Face and OpenAI. Version 1.26.2 reportedly added access to GitHub Actions identity tokens.
Security firms including Socket, StepSecurity, Wiz, JFrog and Endor Labs analyzed the releases. The GitHub Advisory Database rates the incident critical. Its guidance, as relayed in the source material, is to treat any environment that installed either release as exposed, rebuild it using a clean base and rotate credentials that environment could reach, including CI and GitHub Actions secrets.
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 Cloud and Build Credentials
The incident matters because the package was the official release referenced by Microsoft’s installation instructions, rather than an unrelated lookalike package. The source article cites Wiz’s estimate of almost 200,000 downloads a month, though that figure does not establish how many people installed the two affected versions. A malicious release in a trusted software supply chain can reach developer laptops, build servers, containers and automated pipelines through ordinary installation or update processes.
Credentials on those systems can provide access beyond an individual workstation. Cloud keys, source-control tokens and CI secrets may be able to reach code, infrastructure or deployment workflows. The source material does not establish that any stolen credential was used, but the possibility is why researchers and the advisory recommend rotating accessible secrets rather than relying only on removing the package. Short-lived identity tokens also can pose a risk if captured while valid, according to researchers quoted by The Register.
The sequence of uploads highlights a separate issue for registry response: quarantining a malicious version does not necessarily disable the credential used to publish it. StepSecurity says the second release appeared two minutes after the first was quarantined. JFrog likewise noted that removing one version does not by itself revoke compromised publishing access. The timeline underscores why incident response may need to address the publisher account or credential as well as the individual package files.
Products Worth Considering

DJI Mini 5 Pro Fly More Combo with DJI RC 2, 1-Inch CMOS

Anker SOLIX C300 Portable Power Station, 288Wh

Jackery HomePower 3600 Plus Portable Power Station, 3584Wh, 3600W
As an affiliate, we earn on qualifying purchases.
How the Two Releases Appeared
Microsoft’s Agent Framework is an open-source Python toolkit for building AI agents and coordinating workflows. The PyPI package agent-framework acts as an umbrella package, bringing in the core framework and integrations. The source article says version 1.26.0 was published on November 12 and was the last clean release identified by the advisory.
According to the timeline assembled by researchers, version 1.26.1 was uploaded as a wheel file on November 17. After PyPI quarantined it, version 1.26.2 was uploaded under renamed files with additional capabilities, then quarantined eight minutes later. The OpenSSF malware database listed the incident as MAL-2026-20101, and the GitHub Advisory Database published advisory GHSA-764m-3ppc-5m73.
Wiz said it reported its findings to Microsoft on November 17. The source article reports that Microsoft told The Register: “We have made some repositories private while we investigate malicious content.” It also said it had not found a Microsoft security advisory at the time of publication. These statements reflect the source article’s reporting as of November 19, 2026.
Victim Count and Credential Misuse
The number of installations of versions 1.26.1 and 1.26.2 has not been established in the supplied reporting. Wiz’s estimate of almost 200,000 monthly downloads refers to the package overall; it is not a count of installs during the malicious release window or of affected systems. Researchers have not reported how many credentials, if any, were successfully collected.
It is also unclear whether any stolen credentials were used to access accounts, repositories or cloud systems. The source article says StepSecurity found no remote-code-loading second stage in the code it inspected and reports that the malware was not persistent beyond the installed package. Those findings do not determine whether credentials were exfiltrated or whether affected systems were accessed afterward. The full scope of Microsoft’s investigation and any publishing-account compromise also remain unclear in the source material.
Rebuilds and Secret Rotation
Organizations and developers should check package and environment records for installations of agent-framework 1.26.1 or 1.26.2. The advisory’s recommended response, as described in the source article, is to rebuild affected environments from a clean base using version 1.26.0 and rotate every credential those environments could access. This includes secrets held by CI systems and GitHub Actions, not just credentials stored in a developer’s main configuration files.
Teams should also review relevant account, pipeline and cloud activity for unusual access, following their incident-response procedures and the detailed advisories. The supplied article says those advisories contain technical details for defenders but deliberately omits indicators of compromise and attack instructions. Further updates may clarify the number of affected installations, Microsoft’s investigation findings, and whether any credentials were misused; those details had not been established in the source material.
Key Questions
Which package versions were affected?
The affected PyPI releases were agent-framework 1.26.1 and 1.26.2. The GitHub Advisory Database identifies 1.26.0 as the preceding clean release.
Could the code run without using the framework?
Yes. The advisory says a Python start-up hook could run when Python launched in an affected environment, even without importing agent_framework. The framework itself continued to work, according to the source material.
What should someone do if they installed an affected version?
Treat the environment as exposed, rebuild it from a clean base using a clean package version, and rotate credentials it could access. The advisory guidance cited in the source article specifically includes CI and GitHub Actions secrets.
How many users were affected?
The number is not known from the supplied reporting. Wiz’s estimate of almost 200,000 downloads a month applies to the package generally, not to installations of the two malicious versions.
Has Microsoft confirmed credential theft or misuse?
The source article reports Microsoft’s statement that it had made some repositories private while investigating malicious content. It does not report a confirmed victim count, confirmed credential misuse, or a detailed Microsoft security advisory.
Source: ThorstenMeyerAI.com
