Skille w Claude Code: zamień prompty w proces | Solo z Maszyną #10

In this episode, we talk about skills in Claude Code—how to stop explaining everything from scratch to AI and start writing down repetitive workflows. We show how a skill differs from a prompt, when it's worth creating one, where to find ready-made skills, and why it's pointless to install everything someone uploaded to GitHub or Skillta. We also go through specific examples: Skill Creator, Superpowers, auto research, CloudMan, Frontend Designer, NotebookLM, Obsidian Skills, Google Workspace CLI, FFmpeg, Codex Plugin CC, and the start/end session skills Bartek uses in his daily work. This isn't an episode about collecting tools. It's a conversation about how to transform AI from a random chat into a workflow that understands your process, your standards, and can repeat tasks in a similar way. 🎯 Test your AI level. Visit 👉 https://solozmaszyna.pl/, take a quick quiz, and find out what stage of AI implementation you're at in your business. Finally, download the free ebook "The AI ​​Path for Your Business" – a 7-step roadmap that will show you exactly what to do next. A podcast about how to actually use AI at work. No theory. No hype. For solopreneurs, creators, and freelancers. 🎯 Test your AI level – for free! Visit 👉 https://solozmaszyna.pl/ Answer the quiz questions Find out what stage of AI implementation you're at in your business Download the free ebook "The AI ​​Path for Your Business" - a 7-step map that will show you exactly what to do next 🌐 solozmaszyna.pl 👋 BARTEK BĄKOWSKI Instagram:   / bartbakowski   Website: https://bartbakowski.com/pl 👋 JACEK FIEDZIUKIEWICZ Instagram:   / jaca.ai   Website: https://jacekfiedziukiewicz.pl Table of Contents 00:00 - Introduction 00:31 - What's this episode about: skills in Claude Code 01:27 - Why skills are important when working with AI 02:05 - Prompt as a one-time use command 03:37 - Skill as a procedure and repeatable process 05:17 - Prompt vs. skill 06:17 - Skill as an employee skill 08:27 - GitHub as a place to look for skills 09:29 - Skills that update during work 11:35 - First the process, then the skill 12:48 - Skill as a breakdown of work into small steps 14:19 - Let's move on to specific examples 15:00 - Skill Creator 20:42 - How to use Skill Creator in practice 21:07 - Why it's not worth relying on one tool 23:12 - Superpowers 24:18 - Bartek's auto research 26:27 - Feedback loop and AI work checking 29:22 - Skill installation and marketplace 29:52 - CloudMan and dividing tasks into agents 32:11 - Frontend Designer 34:00 - Why you need to match skills to Myself 35:42 - NotebookLM and Working with Sources 38:58 - Obsidian as a Knowledge Base for AI 40:44 - Obsidian Skills 42:12 - Google Workspace CLI 43:20 - FFmpeg for Audio and Video 45:15 - Codex Plugin CC 46:00 - Skillta as a Skill Search Engine 47:52 - Official Skills and Reviews 50:35 - Auto Research, Bartek's Start and End Sessions 53:32 - Skills for Recording Knowledge and Creating Content 53:59 - Summary: When is it Worth Creating Skills? 54:49 - Request for Comments and Skill Recommendations 58:18 - Quiz and Conclusion Hashtags #AI #SztucznaInteligencja #AIwBiznesie #Automatyzacja #Podcast #Claude #ClaudeCode #Skills #SkilleAI #Codex #NotebookLM #Obsidian #solozmachine