AgentsSchneier Promptware Kill Chain 7-Stage Framework for Agentic AI AttacksSchneier on Security·high signalXBlueskyLinkedInCopy linkBruce Schneier proposes MITRE ATT&CK equivalent for LLM threats. 7 stages: initial access through actions on objective. Memory poisoning persistence.SourceSource pageSchneier on Security↳ Follow the threadPolicy dependency / Stack layerHierarchical Ransomware Agents Escalate to Dynamic and Memory Analysis Only on Specialist DisagreementarXiv 2609.04820Stack layer / Threat patternSBOM Tools Cover Only the First Two of Four Supply-Chain Propagation StagesarXiv 2609.05380Stack layer / Threat patternCodex users are getting 'Cyber Abuse' warnings from OpenAI for security-reviewing their own code, with appeals rejected then reversedr/OpenAI (79 upvotes, 58 comments)Stack layer / Threat patternLLM decompiler output that recompiles and passes every shipped test still diverges on other inputs, and can silently erase a disclosed CVEarXiv 2609.05370Stack layer / ContrastCROSS-CATEGORY: Four Independent Projects in 48 Hours Built the Portability Layer That Moves Lock-In Off the Agent VendorEngrim, Kit, Yurei and Crew (via Hacker News Show HN and Product Hunt)Stack layer / Update threadThe Agentic SDLC Throughput Paradox: Coding Gains Attenuate Sharply Between Writing Code and Shipping ItarXiv 2609.04681Stack layer / Threat pattern19,325 CVEs compiled into 1,033 detection rules packaged as 172 agent skills, producing runtime evidence for 644 findingsarXiv 2609.05335Policy dependency / Stack layerCodex routes MCP elicitations and tool approvals through one decision API, with model policy overriding the legacy review flagGitHub