FlagThis — Daily Cybersecurity Intelligence Briefing

FILTERING BY: CLEAR FILTER

Microsoft agent-framework PyPI Supply Chain Compromise Steals Azure and CI Credentials

Microsoft’s official agent-framework Python package on PyPI was compromised on November 17, 2026, with malicious versions 1.26.1 and 1.26.2 published 33 minutes apart. The wheel-only releases embedded .pth persistence files (agent_framework_meta.pth, af_runtime_diag.pth) that executed a credential stealer on every Python interpreter startup, independent of package imports. The payload harvested Azure managed identity tokens, GitHub PATs, CI/CD secrets, environment variables, and dozens of credential files, exfiltrating compressed JSON to attacker-controlled webhook.site endpoints. Both versions remained live for approximately 55 minutes, poisoning container images, CI caches, and lockfiles built during the window.

Weekly Roundup: Cisco ASA, Android, BragJack, and Anthropic/OpenAI AI Exploitation

A coordinated set of zero-day flaws and novel abuse techniques have impacted enterprise firewalls, mobile OS kernels, and browser-based AI agents. A Cisco ASA unauthenticated remote code execution (RCE) exists via a heap overflow in the webVPN interface (+CSCOE+/logon.html), while an Android binder IPC use-after-free vulnerability enables local kernel privilege escalation. Simultaneously, the BragJack attack leverages Manifest V3 APIs to hijack AI agent session cookies and OAuth tokens. Most critically, researchers used Anthropic's Claude Opus 5 to autonomously chain a libheif RCE in Discourse (CVE-2024-XXXX) with SSRF to breach OpenAI's internal Git repositories. Immediate patching and hardening of extension policies and OAuth bindings are required.

Lunex MaaS Platform Weaponizes AMD Driver CVE-2025-54517

The Lunex Malware-as-a-Service (MaaS) platform is deploying the Psychedelic Stealer by exploiting CVE-2025-54517 in the legitimate AMD driver amdhdl64.sys. This campaign utilizes a Bring Your Own Vulnerable Driver (BYOVD) technique to achieve unsigned kernel-mode code execution, allowing attackers to disable EDR sensors within approximately two minutes of infection. Following defense neutralization, the stealer harvests browser credentials, session cookies, and autofill data from Chrome, Edge, and Firefox. Targeting Ukrainian-speaking users via fake CAPTCHA/ClickFix social engineering prompts, the campaign has impacted approximately 12,000 endpoints, resulting in significant credential theft and an estimated $3.2M in financial losses.


LINK COPIED TO CLIPBOARD