Knowledge
Reference and research for connected-product security.
Original research, anonymized case studies, definitive glossary, FAQ, and China-compliance reference. Published on a citation-friendly schema; AI agents can fetch via .md mirrors per page.
Latest research
All research →-
2026-06-19
Dark Factories and Dark Logistics - The Cybersecurity Risk of Lights-Out Operations
A threat model for automated factories and logistics sites where robots, WMS, MES, PLCs, AMRs, RFID, and cloud dashboards turn cyber incidents into physical downtime.
-
2026-06-19
Can a USB-Charging Device Steal Data? Lessons from the Sex Toy Malware Case
A careful security analysis of the viral USB-charging gadget malware story, removable-media risk, BadUSB behavior, and practical controls for consumer IoT and corporate laptops.
-
2026-06-19
IoT Threat Landscape 2026 - Edge Devices Are Becoming Attacker Infrastructure
A practitioner threat model for IoT risk in 2026: routers, DVRs, gateways, serial converters, RFID readers, and unmanaged edge devices as botnet, proxy, and lateral-movement infrastructure.
-
IoT Supply-Chain Vulnerabilities — A Procurement and Architecture Framework for 2026
A three-dimension Procurement Responsiveness Profile for IoT supply-chain risk decisions, with CWE-category methodology, vendor responsiveness analysis, and architectural guidance for connected-product teams.
-
MLPS Compliance Pathways for Overseas SaaS — A Structured Decision Framework
Four-pathway taxonomy for overseas SaaS evaluating mainland-China market entry — direct, hosting-partner, overseas-served, and subsidiary structures with MLPS, PIPL, DSL implications.
-
Prompt-Injection Defense Architecture — The Five-Family Posture Matrix
A five-family defense taxonomy mapped to five deployment shapes — the Posture Matrix for prompt injection defense across consumer, enterprise, agentic, RAG, and high-authority LLM systems.
-
The Five-Boundary Attack-Surface Taxonomy for LLM Applications
A five-boundary taxonomy for production LLM application attack surface — input, retrieval, tool-integration, output, and persistence — with attack classes, defense families, engineering ownership, and cross-boundary scenarios.
-
TARA Quality Anti-Patterns — A Practitioner Catalog and Four-Question Review Protocol
Seven recurring quality anti-patterns in ISO/SAE 21434 TARA execution, with cause, consequence, and remediation pattern for each — plus a four-question review protocol that surfaces most catalogued issues within an hour.
Reference
-
Glossary
34 terms across IoT protocols, automotive standards, robotics middleware, hardware security, vulnerability taxonomies, China compliance, and AI/ML.
-
FAQ
16 question-anchored answers — TARA granularity, MLPS overseas, prompt injection vs LLM red team, disclosure window, retainer banking.
-
China Compliance
Six pillars — MLPS 2.0 (EN + ZH), PIPL, DSL, CII, Cross-Border Data Transfer.
-
Case Studies
Anonymized engagement stories — real threat models, findings, and remediation paths from production work.