A curated set of skills I use with my agents. Feel free to grab what's useful.
Each skill is a small package of instructions and code that gives an agent a new ability — like fixing failing CI pipelines, exploring problems from multiple angles, or cleaning up messy code. This repo is updated as I build and refine new skills.
git clone https://github.com/zvadaadam/az-skills.git
cd az-skills
./scripts/install.shThis connects the skills to your agent. You only need to do this once.
Pull the latest and you're done — new and improved skills load automatically:
cd az-skills
git pull./scripts/uninstall.sh- code-review — Multi-lens code review with 3 parallel sub-agents (correctness, security, design) that validates and reports only high-signal findings
- devs-roundtable — 5 legendary engineers (Carmack, Hickey, Metz, Torvalds, Beck) debate your problem in parallel, then build consensus
- code-simplifier — Reviews code for clarity and maintainability, then cleans it up
- deslop — Detects and removes AI-generated code slop (unnecessary abstractions, over-engineering, verbose patterns)
- design-roundtable — 5 legendary designers (Rams, Ive, Vignelli, Fukasawa, Jongerius) debate your brief in parallel, then build consensus
- brand-name-explore — Generates product/company names using multiple creative personas (Lexicon methodology, poet, linguist, culture hacker, futurist)
- greenlight-pr — Takes a PR, fixes CI failures, addresses review comments, and iterates until everything passes
- interview-me — Interviews you about a plan or design until it has all the context to build the right thing
- ai-journal — Observes how you work with AI and documents patterns, habits, and improvement ideas