Consequential, Not Just Useful
When MCP connects models to permissions, words become actions
Tag: Security
Threat Modeling in the Small
One feature, one flow, one safer change this week
Security Findings as Design Feedback
Reading vulnerabilities as architecture signals
Assumption-Led Security Reviews
Testing claims instead of running checklists
Format String Attacks
From %08x reads to %n arbitrary writes
Buffer Overflow 101
Stack control, step by step, on a deliberately unsafe binary