Most headlines call this a Microsoft story. For a small team, I think it is a lockfile story.
Wiz reported that two malicious versions of Microsoft's agent-framework package, 1.26.1 and 1.26.2, were published to PyPI on 17 November. The Hacker News reported the same compromise. By Wiz's timings, 1.26.1 went up at 20:35 UTC, 1.26.2 at about 21:08 UTC, and both were unavailable by 21:30 UTC. Wiz says: "The most recent safe version is 1.26.0."
Under an hour sounds harmless. It is not, if your build server installs the newest version on every run.
According to Wiz, the code hunted for environment variables with names such as TOKEN, SECRET, PASSWORD, KEY, AZURE and GITHUB. It also went after files like .env, .npmrc, .git-credentials, Docker config and Hugging Face tokens. Wiz adds that the attacker appears to have compromised a client certificate and private key for Microsoft's publishing service. How that happened is not yet known, and I will not guess.
The usual advice is to trust the big vendors. My take is the opposite. Do not trust anyone with your newest version. A package from a giant is still a file that arrived twenty minutes ago.
Here is what I would do today. First, search your lockfiles, CI logs and container images for 1.26.1 or 1.26.2, as Wiz recommends. If you find either, rotate every secret that was in that environment.
Second, pin versions in a lockfile, so a new release reaches you only when you choose it. Boring, and it works.
Third, keep fewer long-lived secrets on build machines. Whatever sits in the environment is exactly what this kind of malware reads. Wiz also lists cloud sign-in tokens among the stolen items for some setups, so if your workflows use them, check the related logins too.
When did you last look at what your build server can read?
Written by
Kristjan SaarikTech entrepreneur building AI agents26 pieces · 11 followersFollow Kristjan to see new posts in your feed, on your phone or by email.
More from
Kristjan Saarik →Cursor told Congress it may "phase out" Chinese AI models
Cursor told two House committees it wants to "explore whether it can phase out" Chinese AI models, CNBC reported on Friday. Nothing changes for you today. It is a statement of intent, not a decision.…
Your next software patch is now a map for attackers
Most headlines read this as a story about hackers. I read it as a deadline for small teams. On Monday, Google's Threat Intelligence Group said it found a framework called exploit_agent. VentureBeat r…
United says no to AI agents booking flights. Why small teams care
United Airlines will block AI agents from booking flights for customers. The reason is not the technology. It is the password. TechCrunch's Zack Whittaker reported it on Thursday. United says it will res…