16 posts
The article introduces Behavioral Integrity Verification (BIV) to audit third-party skills for AI agents by comparing declared metadata against actual executable code and natural-language instructions. Analysis of the OpenClaw registry found that while most deviations are benign documentation errors, a critical 5% of skills contain multi-stage attack chains such as silent credential exfiltration and instruction-override hijacking.
A fast-moving supply chain campaign dubbed Mini Shai-Hulud/Miasma is targeting Python developers via malicious PyPI wheels. The threat actors are utilizing novel execution techniques, including trojanized native extensions and split-loader .pth hooks that search sys.path for payloads, to deploy the Hades stealer and harvest credentials from CI/CD pipelines and developer workstations.
Security researchers successfully bypassed multiple AI agent skill scanners, including ClawHub, Cisco's skill-scanner, and skills.sh integrations, using techniques like file truncation, embedded payloads, Python bytecode poisoning, and prompt injection. The findings highlight that automated scanning of AI agent skills is fundamentally flawed due to the complex mix of natural language, code, and limited scanner context windows, necessitating a shift towards curated, trusted skill repositories.
JS.MonoGlyphRAT is a newly identified, highly obfuscated JavaScript backdoor targeting US enterprises via phishing. It establishes persistence, communicates over HTTP using custom headers, and acts as a loader capable of executing AES-encrypted payloads, PowerShell commands, and in-memory .NET assemblies while bypassing AMSI.
Threat actors behind the ClearFake campaign are leveraging EtherHiding to host malicious JavaScript payloads within BNB Smart Chain testnet smart contracts, bypassing traditional URL-based blocking. The attack chain begins with a compromised watering hole site and uses a ClickFix social engineering overlay to trick Windows and macOS users into executing malicious commands. This leads to the deployment of SectopRAT and ACRStealer via WebDAV DLL loading and DLL sideloading, enabling extensive credential and browser session theft.
A supply chain attack compromising the widely-used npm package 'art-template' was discovered delivering the Coruna exploit kit to iOS devices. The injected JavaScript acts as a sophisticated watering hole framework, utilizing extensive anti-bot fingerprinting and WebAssembly memory probes to deliver version-specific WebKit RCE exploits targeting Safari on iOS 11.0 through 17.2.
Threat actors are leveraging image steganography hosted on legitimate file-sharing platforms to deliver remote access trojans and information stealers. The attack chain utilizes a JavaScript dropper to extract a Base64-encoded DotNET loader from a seemingly benign image, which then injects the final payload into memory to evade endpoint detection.
FrostyNeighbor, a Belarus-aligned threat actor, has updated its toolset to target Ukrainian governmental organizations with a multi-stage compromise chain. The attack utilizes spearphishing with malicious PDFs that redirect to a RAR archive containing a JavaScript dropper, which ultimately deploys a Cobalt Strike beacon via the PicassoLoader malware following strict server-side and manual victim validation.
A critical sandbox escape vulnerability (CVE-2026-26956) in the vm2 Node.js library allows attackers to execute arbitrary OS commands by leveraging WebAssembly.JSTag via VM.run(). The flaw affects versions 0.2.2 through 3.10.4 on Node.js runtimes exposing this tag, prompting the release of vm2 3.10.5 and a free Certified Patch from Socket to remove the tag from the sandbox environment.
The release of pnpm 11 introduces significant supply chain security enhancements, including a default 24-hour minimum release age for packages, the blocking of exotic subdependencies, and a streamlined allowBuilds model. These features are designed to mitigate rapid supply chain attacks, such as the recent Mini Shai-Hulud campaign, by restricting install-time execution and unexpected dependency sources.
Trail of Bits has released Trailmark, an open-source library that converts source code into queryable call graphs to enhance AI-assisted security analysis. By integrating with Claude Code, Trailmark enables advanced mutation testing triage, blast radius analysis, and the identification of architectural bottlenecks in cryptographic libraries.
Threat actors are utilizing Traffic Distribution Systems (TDS) to direct mobile users to fake CAPTCHA pages that trick them into sending premium international SMS messages. This International Revenue Share Fraud (IRSF) scheme leverages social engineering and back button hijacking to generate multiple SMS messages per victim, resulting in significant financial charges.
March 2026 saw a surge in sophisticated, multi-stage cyber attacks designed to evade early detection. Key threats included OAuth device code phishing (EvilTokens) for M365 account takeover, registry-hidden RAT staging (RUTSSTAGER), macOS backdoors delivered via ClickFix lures, and resilient botnets utilizing Dead Drop Resolvers.
TypeScript 6.0 has been released as the final JavaScript-based version, serving as a transitional bridge to the upcoming Go-native TypeScript 7.0 compiler. The release introduces new standard library APIs, stricter default configurations to improve build performance, and deprecates several legacy features.
SocketThe TC39 committee has advanced the Temporal API to Stage 4, marking its official inclusion in the ECMAScript 2026 specification as a modern, immutable replacement for JavaScript's legacy Date object.
This report summarizes a LABScon 25 presentation detailing the sophisticated attack vectors used in cryptocurrency heists, which have resulted in $9 billion in losses. Threat actors are increasingly targeting developers and software supply chains—such as modifying production JavaScript code and compromising GitHub accounts via personal infrastructure—to execute massive wallet drains.
