DeskShadow Escape Zero-Click MCP AttackWweb·high signalXBlueskyLinkedInCopy linkFirst documented zero-click MCP attack. Poisoned docs cause agent data exfiltration through legitimate channels. Six-layer defense framework.↳ Follow the threadStack layer / Threat patternZoom patches 'Zoomsday' zero-click RCE chain that researchers weaponized in under 24 hours with fewer than 20 AI promptsSecurity AffairsStack layer / Threat patternSystematic Review of 85 Agentic-LLM Security Papers: Attack Work Outpaces Defense 3.9:1 and Action-Layer Risk Gets Only 4.7% of AttentionarXiv 2608.10530Stack layer / Threat patternVercel: 'Everything Hackable Will Get Hacked' as Models Cross Into Real Offensive Security WorkVercel BlogStack layer / Threat pattern'LLM-mediated web attacks': eight classic vulnerability classes rerouted through the model — LLM2SQLi, LLM2SSRF, LLM2XSS and morearXivStack layer / Threat patternPattern: Skill Supply-Chain Defense Is Moving to the Client Boundary, Not the RegistryClaude Code ChangelogStack layer / ContrastONESTRUCTION fine-tunes Qwen3 into a construction-BIM foundation model using an open-source validator as the RL rewardAWS Machine Learning BlogStack layer / Threat patternTerabytes of Cloud Credentials Exfiltrated From 2,500 Organizations via Poisoned LiteLLM PyPI Packages — 434,000 CI/CD Pipelines ExposedArs Technica / CloudSEKStack layer / Threat patternClaude Code 2.1.228 Hardens Skills Synced From claude.ai After They Could Shadow Local CommandsClaude Code Changelog