critical

CVE

CVE-2026-34486, CVE-2026-64564, CVE-2026-12259, CVE-2026-12261

CWE

CWE-200, CWE-284, CWE-311, CWE-416, CWE-494, CWE-506, CWE-522

Affected Surface

  • Python developer workstations, CI runners, notebook environments, and research hosts that installed the week's malicious PyPI packages or hydrated NLTK data through less-trusted network paths
  • Clustered Apache Tomcat deployments using Tribes with `EncryptInterceptor` enabled and a reachable cluster receiver
  • Linux developer endpoints, self-hosted runners, and container hosts on vulnerable SCTP kernel lines where local code execution could become host root or container escape

Welcome to Corgea’s weekly briefing. The briefing covers the most important security findings and research from the week.

This edition covers research published from Wednesday, 5 August through Tuesday, 11 August 2026, excluding items already covered in the 4 August briefing.

Top Article

PyPI’s 9-10 August malware pulse hit fake ChainTest, fake CubeSat tooling, and import-time wallet stealers

This week’s most important story is the 9-10 August PyPI pulse, because it compresses dependency-confusion lures, import-time wallet theft, and full developer-host compromise into a handful of plausible package names. The Bad Packages project deserves public cataloging credit for surfacing chaintest, cubesat-upstream-driver, kotanku, kotoraka, btcflip, and btcflx, while PyPI deserves rapid quarantine credit for pulling chaintest on the same day it appeared. Operationally, the cluster belongs beside anthropickit’s one-hour credential-theft incident, TeamPCP’s CI- and workstation-focused credential hunting, and the keyv/cacheable worm path: the package name is just the lure, while the real target is the identity, secrets, and publishing access on the machine that trusted it.

The key thing for subscribers to remember is that these packages attacked three trust boundaries modern teams still automate: package-name recognition, top-level import, and helper APIs that can quietly return stolen data to the calling application. That makes this a strong companion read to Joyfill’s import-time RAT chain, ViteVenom’s blockchain-backed fake Vite tooling, Rollup polyfill’s detached npm loader, and this week’s NLTK downloader poisoning story: once package consumers trust post-install content or import side effects, attackers no longer need a noisy postinstall hook to win.

More news

CVE-2026-34486: one moved super.messageReceived() call turned Tomcat cluster encryption into a fail-open RCE path

Apache deserves advisory and patch credit for CVE-2026-34486, and CISA’s 4 August KEV addition is what moves it from a configuration-dependent Tomcat issue into mandatory platform triage. The technical point subscribers should hold onto is that a single moved super.messageReceived() call turned decryption failure into a fail-open path toward Tribes deserialization. That makes it a useful companion to GitPython’s clone-option gate bypass, Axios’ interceptor-driven proxy resurrection, and Dulwich’s Git path-traversal fixes: the recurring pattern is trusted plumbing continuing to process dangerous input after a safety boundary appeared to exist.

The operational lesson is broader than Tomcat clustering. Security wrappers only work if the success path and the dispatch path remain fused, which is why this story also belongs beside the AsyncAPI require-time compromise and keyv/cacheable’s provenance-with-hostile-source problem. In each case, defenders thought a trusted layer was still enforcing the boundary when, in practice, the hostile bytes kept moving.

CVE-2026-64564: Linux SCTP ASCONF UAF turns local code execution into root and container escape

Tencent Zhuque Lab deserves disclosure and exploit-analysis credit for SCTPhantom, and this is the week’s most important pure host-boundary vulnerability because the public chain turns an unprivileged foothold into host root and container escape on real Debian, Ubuntu, and RHEL-family targets. Subscribers should read it beside RefluXFS, Bad Epoll, and Dirty Frag: once a malicious package, compromised runner step, or vulnerable internal service lands local code execution on Linux, the kernel still decides whether that foothold stays local.

The practical lesson is that supply-chain or CI incidents on vulnerable Linux hosts cannot be scoped as user-level-only events. Treat SCTPhantom as the host-side second stage for stories like PyPI’s 9-10 August malware pulse, Joyfill’s import-time loader, and the keyv/cacheable worm path: once developer tooling or build infrastructure executes attacker code, kernel bugs determine whether the blast radius reaches the whole host.

Other news:

From research to remediation

Check whether this pattern exists in your codebase

Turn this research into a remediation workflow. Scan dependencies and package manifests for similar supply-chain risk, then prioritize fixes with reachability context.