Every week, our malware detection systems scan thousands of new and updated packages across public registries like npm and PyPI. This week was no exception. We confirmed 165 malicious packages between July 17 and July 24, 2026, across npm and PyPI, led by one high-velocity campaign and several new impersonation clusters.
最大的單次事件是 bingo-ai on PyPI, again: nearly 85 versions confirmed in a 23-minute window on July 21, the same automated publishing pattern we flagged in this package last week.
We also confirmed further activity from gcli-control on PyPI, a package our research team profiled in detail this week: a fully-featured Windows RAT (keylogging, webcam/mic capture, clipboard monitoring, browser-credential theft, persistence) that routes its command-and-control through npoint.io, a free JSON-hosting service, rather than an attacker-owned domain. The initial burst (versions 0.1.0 to 0.7.1) escalated to full capability in about 13.5 hours on July 21–22. This week’s digest shows the campaign didn’t stop there: five more versions (0.8.0 through 0.12.0) followed at a steadier pace through July 23–24.
We also confirmed a nine-package Twilio impersonation cluster on npm (twilio-serverless, twilio-functions, twilio-internal, and others), all published July 21 with inflated version numbers, a classic dependency-confusion pattern. A second cluster targeted crypto/DeFi tooling on npm, nine packages including uniswap-sdk-v4, wagmi-react以及 ethers-secure, published July 23.
Smaller but notable: a repeating “date formatting utility” naming theme across three separate campaigns, and n8n-nodes-pwn, a suspiciously named automation-node package confirmed July 22.
這份每週簡報是我們持續進行的工作的一部分。 惡意程式碼摘要我們在此驗證新威脅並提供可操作的情報，以協助 DevSecOps 團隊保護其安全。 pipeline在損失發生之前。讓我們來分析一下本週的發現及其重要性。
165+個包裹。一周內。包裹數量持續攀升。
本週的摘要反映了同樣的趨勢：攻擊者發布的速度比註冊表刪除的速度更快，也比每週一次的掃描能夠偵測到的速度更快。一個 PyPI 包， bingo-ai, went from first version to nearly 85 confirmed versions in about 23 minutes, automated publishing at a speed no manual review process can match. Meanwhile, gcli-control shows that volume isn’t the only risk: a fully-featured RAT that didn’t bother hiding what it was, instead relaying its commands through npoint.io, a free JSON-hosting service, so its traffic blends into ordinary developer activity with no attacker-owned domain to block. And the Twilio and crypto/DeFi impersonation clusters show a third pattern entirely: coordinated, multi-package drops using inflated version numbers to win a dependency-confusion race against internal package names.
Xygeni 早期惡意軟體警告 monitors npm, PyPI, and other registries in real time, flagging threats at the moment of publication, before they reach a build, before an AI agent installs them autonomously, and before a covert relay or a dependency-confusion package has a chance to execute. When bingo-ai ships dozens of versions in under half an hour, or gcli-control routes its C2 through a service your egress rules already allow, detection that runs after the fact is already too late.
Xygeni 的 Open Source Security 該平台為 DevSecOps 團隊提供即時檢測和優先排序功能，幫助他們應對協調供應鏈帶來的壓力，從而保障您的安全。 pipeline保持清潔，同時不拖慢團隊速度。
