Security1 publisher2 min readPublished
Paul McCarty names Pollen Rider as malware that reinfects developers through their own repositories
The public account is a single Risky Business Features episode from September 11, 2026, in which OpenSourceMalware founder Paul McCarty describes how researchers pull malicious packages apart and what one family does after removal.
The Watch · Security desk

What happened
- Risky Business published a Features Podcast episode titled "Hunting software supply chain malware" on September 11, 2026, with OpenSourceMalware founder Paul McCarty and host James Wilson.
- The same episode examines DPRK tradecraft and blockchain-based payload delivery, with malicious packages, GitHub repositories and developer tools as the material under study.
- The pair also discuss LLMs cutting the effort of malware development while introducing operational security mistakes.
Compiled by The WatchSomething wrong?How this is made
Why it matters
- exposure Anyone consuming what an infected developer publishes sits inside the loop the episode describes, so the blast radius of one workstation is that developer's downstream users.
- decision Teams have to decide whether package review and credential rotation run as standing build controls or stay incident tasks; with a reinfection path in the developer's own code, the second choice buys a return visit.
- capability LLM assistance lowers the cost of producing a new malicious package, and the operational security errors it leaves behind are what makes clustering those packages possible.
- constraint A tradecraft interview yields no indicators, so nothing in this account converts into a detection rule or a blocklist entry.
A developer's workstation is also a publishing station, and that is the part of a clean-up that tends to go unchecked. Risky Business's description of the episode says the family McCarty calls Pollen Rider can reinfect developers through their own repositories [7]. Removing the malicious dependency and rotating the tokens does not help if the code that puts it back sits in a repository the developer owns.
The description does not say how the reinfection works. Committed source, a build step, or an artifact the developer publishes and then reinstalls are all consistent with the summary [8]. Which one it is decides which control catches it.
The analysis method is the other half of the episode. McCarty walks James Wilson through static analysis, deobfuscation and the reconstruction of multi-stage kill chains, and the point of the reconstruction is to work out what the attacker is trying to steal [4]. Multi-stage is the operative word for triage. Deobfuscating the first payload tells you how it runs.
LLMs came up in both directions: easier malware development, and operational security mistakes in the result [5]. Cheaper production means more malicious packages to sort through, and the mistakes are what lets a researcher cluster them.
The episode also examines DPRK tradecraft and blockchain-based payload delivery [6], with the material under study being malicious packages, GitHub repositories and developer tools [3]. Those threads stay separate here. The published description does not connect Pollen Rider to DPRK, and it does not say the reinfection behaviour is common across the families McCarty tracks [8].
That is the limit of what a defender can act on. The public description gives no package name, ecosystem, victim count or first-observation date [8], so nothing here becomes a detection rule. What the thin evidence supports is a change in the scope of the response. If a developer's own repositories are a reinfection path, the response to one malicious package includes reviewing what that developer published after the first install, and rotating the token is one step in that work [7].
What to watch
- Package names, ecosystems or a registry advisory that put Pollen Rider on the record beyond this interview.
- Any takedown or registry action tied to reinfection through developer-owned repositories rather than a single malicious release.
- Whether other researchers attribute Pollen Rider to a named actor, which the episode description does not do.