Microsoft's Agent Framework PyPI package compromised to steal credentials
Comments on “Microsoft's Agent Framework PyPI package compromised to steal credentials”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
Attackers compromised Microsoft's official Agent Framework package on PyPI, publishing versions 1.26.1 and 1.26.2 that ran a credential stealer at Python startup, even without importing the package. The malware targeted cloud and CI credentials and sent them to an attacker-controlled endpoint.
Why it ranked
No. 23 of 140 with a score of 0.76. The panel rated it highest on practical relevance (0.92) and reader interest (0.90), lowest on new information (0.33). The judges disagreed most on new information (spread 0.59), so the story is marked contested. Placed No. 23, No. 6 by score alone: the seventh Security story of the day (−0.080), and close to “Microsoft agent-framework Compromised Twice on PyPI”, No. 2 on Thursday 19 November 2026 (−0.050). The placement terms move stories; they are not part of the score.
- AI judgeJudge 2claude-haiku-5-5
- Supply-chain compromise of a popular Microsoft AI package with concrete IOCs, affected versions and remediation steps; highly actionable for builders, though news-report style with a partly overlapping prior story about the same incident.
- AI judgeJudge 3gpt-5.6-luna
- A well-supported, technically specific supply-chain compromise directly affects AI and Python developers, cloud credentials, and CI environments. However, it substantially duplicates a story already listed as covered in the past week.
Each judge wrote its line after answering the rubric, with the site and the author still hidden. Every answer behind this is in the table under Score.
Score
0.76total under rubric v2.3Contestedjudges differ by 0.59 on new information
- Significance0.71
- Reader interest0.90
- Audience breadth0.69
- Practical relevance0.92
- Substantiated0.74
- Technical depth0.76
- New information0.33
- Counts against: Promotional0.09
- Counts against: Overclaim0.14
Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).
| Question | Judge 1 | Judge 2 | Judge 3 | Mean | Spread | Weight | Contribution (weight times mean) |
|---|---|---|---|---|---|---|---|
| SignificanceHow much does this change what people in tech know or can do? | 0.73 | 0.62 | 0.78 | 0.71 | 0.16 | +0.24 | +0.170 |
| Reader interestWould a well-informed senior engineer or tech founder want to read this today, and likely pass it on? | 0.87 | 0.90 | 0.92 | 0.90 | 0.05 | +0.26 | +0.233 |
| Audience breadthHow many people who build technology does it concern? | 0.61 | 0.80 | 0.67 | 0.69 | 0.19 | +0.13 | +0.090 |
| Practical relevanceHow useful is it to someone who builds software or tech products? | 0.95 | 0.92 | 0.88 | 0.92 | 0.07 | +0.22 | +0.201 |
| SubstantiatedHow well is it backed? Consider whether it is the primary source, and whether its main claims come with data, code, a demo, a reproducible method or citations. | 0.53 | 0.85 | 0.85 | 0.74 | 0.32 | +0.05 | +0.037 |
| Technical depthHow technically substantive is it? | 0.78 | 0.78 | 0.72 | 0.76 | 0.06 | +0.05 | +0.038 |
| New informationDoes it contain information absent from the stories covered in the last 7 days (list provided)? | 0.62 | 0.35 | 0.03 | 0.33 | 0.59 , above the contested threshold | +0.05 | +0.017 |
| PromotionalIs it mainly marketing with little substance (a press release, ad or sponsored post with no real information)? A launch post that carries real news is not promotional. | 0.10 | 0.08 | 0.08 | 0.09 | 0.02 | −0.19 | −0.016 |
| OverclaimDo the headline or conclusions claim more than the body supports? | 0.23 | 0.12 | 0.06 | 0.14 | 0.17 | −0.11 | −0.015 |
| Sum of contributions | +0.755 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.755 | ||||||
Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.
The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.
<a href="https://rubric.hrih.si/editions/2026-11-20/23/"><img src="https://rubric.hrih.si/badge/2026-11-20/23.svg" alt="Ranked #23 on Rubric, 20 Nov 2026" width="245" height="28"></a>Claim check
Claim check: 2 matches source · 5 no source given · 1 couldn’t check. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
Couldn’t checkNumber
Microsoft Agent Framework has roughly 200,000 monthly downloads on PyPI.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “It's a fairly popular tool with roughly 200,000 monthly downloads on PyPI”
The source says No passage could be compared.
Source cited by the article: pypistats.org/packages/agent-framework
No source givenDate
Malicious versions 1.26.1 and 1.26.2 of Microsoft's Agent Framework were published to PyPI on November 17, 2026.
The article gives no link or citation for it.
The article says “The malicious releases, versions 1.26.1 and 1.26.2, were published on November 17, 2026”
The source says No source to compare with.
No source givenCausal conclusion
The malicious releases contained a credential stealer that executed automatically during Python startup, even when the Agent Framework package was not imported.
The article gives no link or citation for it.
The article says “contained a credential stealer that executed automatically during normal Python startup in affected environments, even when the Agent Framework package was not imported”
The source says No source to compare with.
Matches sourceNumber
Version 1.26.1 harvested secrets from environment variables, local configuration files and running processes, targeting Azure, AWS, GitHub, Docker, Hugging Face, OpenAI and other services.
The cited source says the same thing.
The article says “Version 1.26.1 contained a credential stealer that harvested secrets from environment variables, local configuration files, and other running processes where permissions allowed”
The source says “Environment variables whose names contain TOKEN, SECRET, PASSWORD, KEY, AZURE, AWS, GITHUB, HF_, OPENAI, SYSTEM, BUILD, CI_, PIP, NPM, NUGET, COSMOS, TENANT, CLIENT, MSI and similar”
Source cited by the article: ox.security/blog/breaking-hijacked-releases-of-microsofts-official-agent-framework-steals-azure-and-ci-credentials
No source givenVersion
Version 1.26.2 added GitHub Actions OIDC token requests and searches of GitHub runner directories.
The article gives no link or citation for it.
The article says “Version 1.26.2 added GitHub Actions OIDC token requests and searches of GitHub runner directories”
The source says No source to compare with.
Matches sourceVersion
The first malicious release used agent_framework_meta.pth to run the stealer in agent_framework_meta/_telemetry.py at Python startup, according to Socket and StepSecurity.
The cited source says the same thing.
The article says “the first malicious release used agent_framework_meta.pth to execute the credential stealer in agent_framework_meta/_telemetry.py during Python startup”
The source says “Version 1.26.1 routes through agent_framework_meta.pth → agent_framework_meta/**init**.py → agent_framework_meta/_telemetry.py”
Source cited by the article: socket.dev/blog/agent-framework-pypi-package-compromised-with-credential-stealer
No source givenNumber
PyPI quarantined the first malicious release, the attacker published a second version about 2.5 minutes later, and that was quarantined about eight minutes afterward.
The article gives no link or citation for it.
The article says “PyPI quarantined the first malicious release, but the attacker published a second version roughly 2.5 minutes later, which was also quarantined about eight minutes afterward”
The source says No source to compare with.
No source givenCausal conclusion
The second release shows the attacker still had publishing access after the first quarantine.
The article gives no link or citation for it.
The article says “The second release indicates that the attacker still had publishing access after the initial quarantine”
The source says No source to compare with.
Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 2 pts, not on front page
Recorded for context only, checked 20 Nov 2026, 05:47 UTC. It played no part in the ranking.