Hey Everyone!

I've been working on something I want to share with you. As you know, I've started relying on AI assistants to do most of my coding, but I review everything that is being generated.

I've grown frustrated by the slop being generated and the amount of corrections I had to do. After years of building with Node.js, Fastify, and TypeScript, I've learned a lot: patterns that work, tools that save time, and gotchas that bite you when you least expect it. I need my AI assistants to match (or at least get close) my expectations.

I decided to organize all of this into a collection my AI assistant can use to help me work faster. Here is my skills repo that you can start using with a simple: npx skills add mcollina/skills. This repo encode my preferences and best practices so I don't have to repeat myself every time. You might find it useful as well.

What's inside?

The repo includes skills for:

fastify

Best practices for Fastify development - hooks lifecycle, plugin architecture, and performance tuning. Based on years of work on Fastify core.

node

Best practices for Node.js development - event loop patterns, async error handling, stream processing, and the Node.js test runner. Common pitfalls and modern patterns from Node.js core.

nodejs-core

Deep internals including C++ addons, V8 internals, libuv patterns, and build systems. For understanding Node.js at the lowest level.

typescript-magician

Advanced type systems, complex generics, and removing any types. Inspired by Matt Pocock at TotalTypeScript - his course is excellent.

octocat

Git and GitHub workflows using the gh CLI. PRs, issues, releases, and repository management.

oauth

OAuth 2.0/2.1 with Fastify integration. Based on RFC 6749 and work on @fastify/oauth2.

linting-neostandard-eslint9

Linting with neostandard and ESLint v9 flat config. Modern linting without the drama.

documentation

Technical writing following the Diátaxis framework - tutorials, how-to guides, reference, and explanation.

Each skill has markdown files, code snippets, and config examples. My AI assistant reads these and applies the patterns when helping with code reviews or writing new features. It's like having a seasoned developer pair program with you - except this one remembers that V8 flag you mentioned in 2022.

What is a Skill?

Skills follow the open Agent Skills standard - started by Anthropic, now supported by many AI agents.

A skill is a folder with:

• SKILL.md - metadata and instructions
• scripts/ - optional executable code
• references/ - optional docs
• assets/ - optional templates

The SKILL.md needs name and description in the frontmatter, then markdown with instructions. Agents load metadata first, then full instructions only when needed.

The standard works with OpenAI Codex, GitHub Copilot, Claude Code, skills.sh, and others. There's even a gaming-focused agent - check out pi coding-agent for something different.

What's next

I'll add more skills as I work on new projects and dive into performance optimization, security, and deployment patterns. If you find this useful, let me know!

Check it out at github.com/mcollina/skills.

Thanks!