Jamie's tech blog

The Blue Nowhere

[AGENT-SECURITY] MCP security (to 202604)

Global minimum view of MCP security

According to multiple papers, make a roughly category of MCP security.

[AGENT-SECURITY] Abount Prompt Injection(to 202604)

Global minimum view of Prompt Injection

According to multiple papers, make a roughly category of prompt injection.

[AGENT-SECURITY] After scanning 3000+ AI Agent Security papers, what could we see?

Global minimum view of AI Agent Security

Going through the AI agent security papers — they roughly fall into a five-layer stack.