40 Threat Scenarios for MCP, A2A and AG-UIA browsable threat library of 40 structured attack scenarios targeting MCP, A2A and AG-UI agent protocols, mapped to MITRE ATT&CK, ATLAS and OWASP.5 min read·olegm.hashnode.devMar 20
Depth vs breadth: the two kinds of AI agent security testingMost agent security tooling focuses on finding new vulnerabilities. But who's checking whether last month's prompt injection still works after a model7 min read·olegm.hashnode.devMar 12
A Declarative Schema for MCP Attacks: Why We Need OneMapping the OWASP MCP Top 10 to Adversarial YAML Definitions9 min read·olegm.hashnode.devFeb 13
