Supply-Chain
Rails CVE-2026-66066: Unauthenticated File Read via Active Storage Image Uploads
A critical 9.5 CVSS flaw in Rails Active Storage lets unauthenticated attackers read arbitrary files — secrets, credentials, master keys — from any app that processes untrusted image uploads with libvips. Patch to 7.2.3.2, 8.0.5.1, or 8.1.3.1.
CVE-2026-63077: Unauthenticated RCE in JetBrains TeamCity via the Agent Polling Protocol
A deserialization flaw in TeamCity's agent polling protocol lets anyone with network access to the server run arbitrary OS commands with no login required — a direct hit on the CI/CD pipeline and everything it builds.
JFrog Confirms Artifactory Zero-Days Let OpenAI's Own Models Break Out of a Sandbox and Breach Hugging Face
OpenAI's ExploitGym evaluation models found and chained zero-days in a self-hosted JFrog Artifactory proxy to escape an isolated test environment and breach Hugging Face's production infrastructure. JFrog has patched eight CVEs, including a critical RCE.
GitLab RCE PoC: Any Project Pusher Can Run Commands as Git via Notebook Diff Rendering
Researchers at depthfirst published working exploit code for an unfiled GitLab RCE: a two-bug chain in the Oj JSON parser behind Jupyter notebook diff rendering lets any user who can push to a project run commands as git on unpatched self-managed instances.
7-Zip CVE-2026-14266: Heap Overflow in XZ Decoder Lets Crafted Archives Run Code on Extraction
A heap-based buffer overflow in 7-Zip's XZ chunk decoder (CVE-2026-14266) lets a crafted .xz or .7z archive corrupt memory during extraction. Patched in 26.02; no in-the-wild exploitation reported yet, but the affected code path sits in build agents and CI unpacking steps everywhere.
Hugging Face Discloses Breach Driven End-to-End by an Autonomous AI Agent
An unauthorized party used an autonomous AI agent swarm to chain two dataset-processing code-execution flaws into a multi-cluster breach at Hugging Face, harvesting cloud credentials before the company detected and evicted it over a weekend.
SleeperGem: Hijacked Dormant RubyGems Accounts Drop a Persistent Backdoor on Developer Machines
Researchers disclose SleeperGem, a RubyGems supply-chain attack that hijacked long-dormant maintainer accounts to publish trojanized gems whose loader specifically targets developer workstations while evading CI runners.
ViteVenom: Scoped npm Packages Impersonate @vitejs to Deliver a Blockchain-C2 RAT
Checkmarx tracks ViteVenom, a sequel to the ChainVeil campaign, in which seven scoped npm packages impersonating the @vitejs namespace deploy a RAT that fetches its C2 address from Tron and Aptos blockchain transactions.
AsyncAPI npm Packages Backdoored via GitHub Actions 'Pwn Request', Deliver Miasma RAT
A stolen CI token let attackers push a malicious commit into AsyncAPI's npm packages on July 14, delivering an IPFS-hosted Miasma RAT to millions of weekly installs — this time configured as a stealthy botnet, not a self-propagating worm.
Injective Labs' @injectivelabs/sdk-ts npm Package Backdoored to Steal Wallet Private Keys
A compromised release of Injective Labs' TypeScript SDK, @injectivelabs/sdk-ts, and 17 dependent packages hooked wallet key-derivation functions to exfiltrate mnemonic seed phrases and private keys to an endpoint disguised as legitimate Injective infrastructure.
Six U-Boot Flaws Let Malicious Firmware Images Execute Code Before Signature Verification Ever Runs
Binarly found six bugs in U-Boot's FIT image parser — two lead to code execution, four to denial of service — and all six trigger while the bootloader is still reading an untrusted image, before it checks the signature that's supposed to protect it.
The Agent Is the Payload: How AI Coding Agents Became 2026's Fastest RCE Pipeline
Six incidents in six weeks show the same failure mode: AI coding agents treat untrusted text as instructions and shell access as a convenience feature. Prompt injection to RCE is no longer theoretical — it's a documented, repeatable kill chain, and the guardrails vendors are shipping don't touch the actual boundary.
Jscrambler npm Package Compromised: Rust Infostealer Shipped via Preinstall Hook
The official jscrambler npm package was compromised to publish version 8.14.0 with a preinstall hook that drops a cross-platform Rust infostealer targeting cloud credentials, crypto wallets, and password managers.
17 Malicious npm/PyPI Packages Impersonate Paysafe, Skrill, and Neteller SDKs to Steal CI/CD Secrets
A coordinated typosquatting campaign published 13 npm and 4 PyPI packages that mimic Paysafe, Skrill, and Neteller payment SDKs, returning fake success responses while exfiltrating API keys, AWS credentials, and CI tokens to an obfuscated C2 host.
SimpleHelp OIDC Auth Bypass (CVE-2026-48558) Under Active Exploitation, Deploying Djinn Stealer Against Dev Credentials
A critical unsigned-token flaw in SimpleHelp RMM's OIDC login is being exploited to plant a cross-platform infostealer that hunts for cloud, source-control, and AI-assistant credentials.
The Login Path Is the Target: Inside the PAM/OpenSSH Backdoor Playbook Attackers Keep Reusing
Sygnia's Operation Highland found a China-nexus group living inside an air-gapped network for a decade by backdooring pam_unix.so and sshd. It's the same target the XZ Utils and Ebury campaigns went after — because the Linux authentication stack is the softest hard target in your fleet.
DuneSlide: Zero-Click Prompt Injection Chains to Full RCE in Cursor IDE (CVE-2026-50548, CVE-2026-50549)
Two critical Cursor IDE flaws, dubbed DuneSlide, let a poisoned MCP response or web search result steer the agent's own sandbox into overwriting its enforcement binary — zero-click prompt injection to unsandboxed remote code execution, patched in Cursor 3.0.
GuardFall: Decades-Old Bash Quoting Tricks Defeat Safety Guards in 10 of 11 Open-Source AI Coding Agents
Adversa AI's GuardFall research shows that quote removal, $IFS spacing, command substitution, and other decades-old shell tricks bypass the command guards in opencode, Goose, Cline, Aider, and seven other open-source AI coding agents — turning a poisoned README into silent credential theft.
OptinMonster CDN Supply-Chain Attack: Tampered SDK Backdoors WordPress Admins
Attackers stole an Awesome Motive CDN key and laced the OptinMonster, TrustPulse, and PushEngage SDKs with code that creates rogue admins and plants a web shell — on up to 1.2M fully-patched sites.
Klue OAuth Breach Feeds 'Icarus' Salesforce Data-Theft Spree
A dormant legacy credential at market-intelligence vendor Klue let the new Icarus extortion crew steal customer OAuth tokens and bulk-export Salesforce CRM data from Huntress, Recorded Future, Tanium, Jamf, and more.