101 Hacking Skills Packaged for AI Agents

What it does

hack-skills is a structured knowledge base that distills public security research into 101 discrete markdown skills across 14 domains—from web app attacks and Active Directory to blockchain and AI/ML security. Instead of raw payload dumps, it compresses methodologies into a three-tier hierarchy: a master router entry, six category entries, and deep topic playbooks meant to be loaded on demand by an AI agent.

The interesting bit

The project ships in three synchronized forms: plain GitHub source, a static fuzzy-searchable web UI, and an AES-256 encrypted ZIP with a public password so markdown can be used offline or behind strict AV. That distribution awareness—treating a knowledge base as software that needs deployment—is unusual for a collection of markdown files.

Key highlights

• 101 skills covering web, API, auth, OS privilege escalation, AD attacks, mobile, binary exploitation, reverse engineering, cryptography, blockchain, AI/ML, networking, and forensics
• Hierarchical loading: one master entry routes to six category entries, which then load specific deep skills only when needed
• Static web UI with weighted fuzzy search, tier filters (P0/P1/P2), and field qualifiers like category:auth or tier:deep
• Distilled from public sources (PayloadsAllTheThings, hacktricks, ctf-wiki, etc.) into small, stable, auditable methodology files rather than raw dictionaries
• Uniform structure: every skill lives at skills/{semantic-identifier}/SKILL.md for predictable programmatic access

Verdict

Worth bookmarking if you are building a security-focused AI agent or LLM toolchain and need a maintainable, routable source of offensive knowledge. Skip it if you are looking for a standalone exploitation framework—this is purely a structured knowledge layer.