Skillsets on the Chain: A Blockchain-based Zero-Trust Framework for Agentic AI Networking

A collection of newly posted cybersecurity and AI-security research papers. Key findings include widespread authentication gaps and exploitable vulnerabilities in internet-facing Model Context Protocol servers, new multi-agent and LLM attack surfaces, defenses against evolving prompt attacks, automated attack-chain generation, healthcare-IoT threat detection, and distributed access-control mechanisms. The MCP assessment reports SQL injection, SSRF to cloud metadata, prompt injection, path traversal, and unauthenticated shell-capable tools at scale.