How I Built an Open-Source Cybersecurity Skills Database for AI Agents (611+ Skills)

How I Built an Open-Source Cybersecurity Skills Database for AI Agents (611+ Skills) AI agents are transforming software engineering. Tools like Claude Code, GitHub Copilot, and Cursor can write code, debug issues, and refactor entire codebases. But ask one to analyze a memory dump from a compromised server, triage a SIEM alert, or assess an Active Directory attack path, and you get generic advice that no security practitioner would follow. I built an open-source database of 611 cybersecurity skills structured for AI agent consumption. This post explains why, how, and what the skills actually look like. The Problem: AI Agents Lack Security Expertise When a security analyst encounters a suspicious process on a compromised Windows host, they don't think in generalities. They immediately: Check the process tree for parent-child anomalies Run vol3 -f memory.dmp windows.malfind to detect injected code Extract suspicious memory regions for YARA scanning Cross-reference process network connec