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 18 of 2026 · Retrospective · Covers 21 April 2026 11:00 to 28 April 2026 11:00 UTC

Read on the web

Top developments

Microsoft releases emergency patches for critical ASP.NET flaw

BleepingComputer · Advisory · Remediation, SCA

CISA orders feds to patch BlueHammer flaw exploited as zero-day

BleepingComputer · Advisory · Remediation

New Checkmarx supply-chain breach affects KICS analysis tool

BleepingComputer · Incident · SCA

New npm supply-chain attack self-spreads to steal auth tokens

BleepingComputer · Incident · SCA

PyPI package with 1.1M monthly downloads hacked to push infostealer

BleepingComputer · Incident · SCA

SAST

GoAT-X: A Graph of Auditing Thoughts for Securing Token Transactions in Cross-Chain Contracts

arXiv · Research · SAST

Evaluating Cryptographic API Misuse Detectors for Go

arXiv · Research · SAST

CrossCommitVuln-Bench: A Dataset of Multi-Commit Python Vulnerabilities Invisible to Per-Commit Static Analysis

arXiv · Research · SAST

Strategic Heterogeneous Multi-Agent Architecture for Cost-Effective Code Vulnerability Detection

arXiv · Research · SAST

SCA

GHSA-6w67-hwm5-92mq / CVE-2026-33626: LMDeploy has Server-Side Request Forgery (SSRF) via Vision-Language Image Loading

GitHub Advisory Database · Advisory · SCA

GlassWorm malware attacks return via 73 OpenVSX "sleeper" extensions

BleepingComputer · Analysis · SCA

Bitwarden CLI npm package compromised to steal developer credentials

BleepingComputer · Incident · SCA

Mythos pulls zero-days forward. Here's what you need to know now.

Chainguard · Analysis · Remediation, SCA

DAST

Automation-Exploit: A Multi-Agent LLM Framework for Adaptive Offensive Security with Digital Twin-Based Risk-Mitigated Exploitation

arXiv · Research · DAST, Models

Agentic SDLC

RAT: RunAnyThing via Fully Automated Environment Configuration

arXiv · Research · Agentic SDLC

Monitoring Claude Code/Cowork at scale with OTel in Elastic

Elastic · Analysis · Agentic SDLC, AI Security

AutoRISE: Agent-Driven Strategy Evolution for Red-Teaming Large Language Models

arXiv · Research · Agentic SDLC, Models

Synthesizing Multi-Agent Harnesses for Vulnerability Discovery

arXiv · Research · Agentic SDLC

Remediation

CVE-2026-32202: Microsoft Windows Protection Mechanism Failure Vulnerability added to CISA's Known Exploited Vulnerabilities catalog

CISA · Advisory · Remediation

Apple fixes bug that let the FBI recover deleted Signal messages

BleepingComputer · Analysis · Remediation

NIST NVD Update: What it Means For Vulnerability Management

Wiz · Standard / policy · Remediation

Microsoft Patches Entra ID Role Flaw That Enabled Service Principal Takeover

The Hacker News · Advisory · Remediation

Trust

C8s: A Confidential Kubernetes Architecture

arXiv · Research · Trust

ArgRE: Formal Argumentation for Conflict Resolution in Multi-Agent Requirements Negotiation

arXiv · Research · Trust

What Should Frontier AI Developers Disclose About Internal Deployments?

arXiv · Research · Trust

Sovereign Agentic Loops: Decoupling AI Reasoning from Execution in Real-World Systems

arXiv · Research · Trust

Models

The Safety-Aware Denoiser for Text Diffusion Models

arXiv · Research · Models

Poisoning Learned Index Structures: Static and Dynamic Adversarial Attacks on ALEX

arXiv · Research · Models

When Prompt Under-Specification Improves Code Correctness: An Exploratory Study of Prompt Wording and Structure Effects on LLM-Based Code Generation

arXiv · Research · Models

DETOUR: A Practical Backdoor Attack against Object Detection

arXiv · Research · Models

AI Security

Making AI-Assisted Grant Evaluation Auditable without Exposing the Model

arXiv · Research · AI Security

AgentWard: A Lifecycle Security Architecture for Autonomous AI Agents

arXiv · Research · AI Security

A Comparative Evaluation of AI Agent Security Guardrails

arXiv · Research · AI Security

Layerwise Convergence Fingerprints for Runtime Misbehavior Detection in Large Language Models

arXiv · Research · AI Security, Models

Advisories in this window

  • 108 npm advisories · 52 critical · 24 high · 26 moderate · 6 low
  • 60 Go advisories · 10 critical · 22 high · 22 moderate · 6 low
  • 45 Maven advisories · 7 critical · 18 high · 19 moderate · 1 low
  • 35 crates.io advisories · 2 critical · 6 high · 20 moderate · 7 low
  • 33 PyPI advisories · 2 critical · 10 high · 16 moderate · 5 low
  • 25 Packagist advisories · 2 critical · 8 high · 10 moderate · 5 low
  • 6 NuGet advisories · 6 moderate
  • 6 RubyGems advisories · 2 critical · 3 high · 1 moderate
  • 2 Multi-ecosystem advisories · 1 critical · 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.