PREVIEW. This is how it will look in an inbox; it was not sent to you. Each real email carries its own unsubscribe link and Keryn's postal address.

Keryn Headlines

Weekly, week 45 of 2026 · Retrospective · Covers 27 October 2026 11:00 to 3 November 2026 12:00 UTC

Read on the web

Top developments

Drupal warns of critical Commerce Core flaw exploited in attacks

BleepingComputer · Incident · Remediation

New Cisco firewall zero-day exploited to crash SSL VPN devices

BleepingComputer · Incident · Remediation

Security notice — malicious version of Anthropic.claude-code (2.1.312 and 2.1.313) identified in the Cursor VS Code Extension Marketplace

Cursor · Incident · Agentic SDLC, AI Security, SCA

CISA says max severity GitLab flaw now exploited in the wild

BleepingComputer · Incident · Remediation

Ivanti patches two more Connect Secure zero-days chained in attacks

BleepingComputer · Advisory · Remediation

SAST

Towards Effective Fuzzing of EDK2 Components

arXiv · Research · DAST, SAST

Fixed Detectors, Moving Code: How Well Do Static Security Analyzers Hold Up over a Decade of C/C++ Evolution?

arXiv · Research · SAST, SCA

How Deterministic SAST and AI Security Agents Divide the Work: Coverage, Harness Policy and Cost in Snyk VulnBench 2.0

arXiv · Research · Agentic SDLC, AI Security, SAST

Structured Security Explanations for Learning-Based Vulnerability Detection

arXiv · Research · SAST

SCA

Anthropic VS code extension on Cursor backdoored for nearly 12 hours

Ars Technica: Security · Incident · Agentic SDLC, AI Security, SCA

A Supply Chain Attack Put Malicious Anthropic Claude Extensions on Thousands of Devices

WIRED · Incident · Agentic SDLC, SCA

Composer 2.11 Adds Dependency Cooldowns

Socket · Release · SCA

Alpine Linux rebuilds GitLab after attempted RCE attack

BleepingComputer · Analysis · SCA

DAST

China-backed threat group used agentic AI pentesting tool to compromise 140 sensitive targets in 11 days

CyberScoop · Incident · DAST

DYNAS: A Dynamic Network Penetration Testing Framework Using Deep Hierarchical Reinforcement Learning

arXiv · Research · DAST

Agentic SDLC

Hugging Face malware campaign returns to hijack GPUs for crypto mining

BleepingComputer · Incident · Agentic SDLC, Models

DoGNAVY-Exploit: Harnessing Parallel Open-Weight Agents for Stage-Aware Vulnerability Exploitation

arXiv · Research · Agentic SDLC

VeriHarm-Bench: Measuring Harmful Success in LLM Coding Agents Verifiably

arXiv · Research · Agentic SDLC, Models

Optimize the Harness, Not the Model: Fitting a Minimal Agent to a Self-Hostable LLM for Investigative Analysis

arXiv · Research · Agentic SDLC

Remediation

Inside KeyDrift: Analyzing historical UAT-11473 activity on Cisco SD-WAN

Cisco · Analysis · Remediation

New Android SMS of Death exploit can remotely crash unpatched phones

BleepingComputer · Research · Remediation

Hackers now scanning systems for recently-patched Oracle flaw

BleepingComputer · Advisory · Remediation

Adobe Fixes 8 Critical Campaign Classic Flaws Allowing Code Execution and Security Bypass

The Hacker News · Incident · Remediation

Trust

Trustworthy Execution of Industrial Agents

arXiv · Research · Trust

New LF Research Survey Shows Generative AI Cuts Open Source Coding Time in Half, but 25% of Incoming Pull Requests are Considered AI Slop

OpenSSF · Analysis · Trust

Agentic process evidence

arXiv · Research · Trust

Omission-Safe Conditional Contracts for Evidence-Closed CI Qualification

arXiv · Research · Trust

Models

BrainwashedLLM: Jailbreaking via Automated Adversarial Chain of Thoughts Discovery

arXiv · Research · Models

The Web Doesn't Sit Still: Adversarial Self-Evolving Attacks on Search Agents

arXiv · Research · Models

Leakage-Aware Encrypted Semantic Communication with Key-Conditioned Signal Perturbation

arXiv · Research · Models

LoRA-Patch: Defending Against Backdoored LoRA Adapters via Directional Steering

arXiv · Research · Models

AI Security

Right Tool, Wrong Parameters: Defending LLM Agents Against Parameter Poisoning

arXiv · Research · AI Security

The Devil is in the Context: Nine Dimensions of Prompt Injection Attacks

arXiv · Research · AI Security

MCP Tool Poisoning in Agentic AI Frameworks: A Cross-Framework Measurement Study

arXiv · Research · AI Security

RuntimeTickets: Source-Grounded Security Obligations for Agent Skills

arXiv · Research · AI Security

Advisories in this window

  • 294 npm advisories · 210 critical · 47 high · 28 moderate · 9 low
  • 104 PyPI advisories · 23 critical · 39 high · 37 moderate · 5 low
  • 33 Go advisories · 1 critical · 13 high · 18 moderate · 1 low
  • 17 Packagist advisories · 11 high · 5 moderate · 1 low
  • 15 Maven advisories · 5 high · 9 moderate · 1 low
  • 15 crates.io advisories · 2 critical · 4 high · 4 moderate · 5 low
  • 9 RubyGems advisories · 4 critical · 1 high · 4 moderate
  • 3 Multi-ecosystem advisories · 1 high · 1 moderate · 1 low
  • 1 GitHub Actions advisory · 1 critical
  • 1 NuGet advisory · 1 moderate

All advisories

Titles, links and short excerpts only; every item links to its source. Selection is automatic; the editor's note is AI-drafted and checked as described. Preview: the unsubscribe link and Keryn's postal address are added to each email when it is sent.