A collection of Claude skills for guidance on Swift (Server) development, along with a linting tool that enforces best practices. Claude, as any LLM (or Human) makes mistakes. When Claude makes a mistake, most people have one of two reactions: - Tell Claude it sucks, and let it try again. - Open up some social media, and vent about how Claude sucks. However, trying the same with Humans has proven
Add this skill
npx mdskills install Joannis/claude-skillsMeta-documentation about a development methodology; lacks actual actionable skill instructions
1# Swift Server Skills & Linting23A collection of Claude skills for guidance on Swift (Server) development, along with a linting tool that enforces best practices.45## The Technique67Claude, as any LLM (or Human) makes mistakes. When Claude makes a mistake, most people have one of two reactions:8- Tell Claude it sucks, and let it try again.9- Open up some social media, and vent about how Claude sucks.1011However, trying the same with Humans has proven extremely ineffective. *Humans bite back.*1213### The solution1415The solution is a pretty simple feedback loop:161. Whenever you generate code, _read it_ before you commit.172. If you see a mistake, try to articulate what should be improved and _why_.183. Iterate 2. until you have a rationale Claude can follow easily.194. Add the rationale with a small code sample to a Claude skill, to ensure future iterations have the same context.2021And now the critical step: **Encode your newfound rule/skill as a linter rule.**2223This last step ensures that future code generated by Claude will follow these same rules, independently of whether the skill is in the LLM's context window.2425## Overview2627This repository contains two main components:28291. **Skills** - Knowledge bases that provide expert guidance on Swift server topics302. **swift-server-lint** - A linting tool that enforces best practices from the skills
Full transparency — inspect the skill content before installing.