New written course: Claude Code & Agent Workflow
Hey friend,
Learning Claude Code properly right now looks like this: one YouTube video for hooks, a Reddit thread for MCP, three docs pages for permissions, and a blog post that was already outdated when you found it.
So I wrote it all down in one place.
24 lessons in 7 modules. Every command, setting, and hook in it was checked against the official docs this month, not copied from a video from last spring.
- Context - install a CLAUDE.md that actually works, and how to keep a session sharp with /context, /compact, and /clear
- Control - plan mode, permission rules in settings.json, and safe undo with /rewind + git
- Hooks - 6 ready recipes: block dangerous commands, protect your secrets, format on every edit, run tests before Claude says "done", and get pinged when it needs you
- Extend - your own slash commands, skills that actually trigger, subagents, plugins
- Tools - MCP set up the way a team would: scopes, config, security
- Automate - Claude running headless in your scripts and in CI (GitHub Actions)
- Daily workflow - the loop I use every day, a team setup, and a one-page cheat sheet
Each lesson ends with a short exercise you do on your own repo. Video versions are coming next. I'm recording them now, and they land right inside each lesson.
One more change: AI for .NET Developers is now AI for Developers
When I started the community, it was for .NET developers only. Then the questions started coming in: "Does this work with my React frontend?", "Can I use the skills in a Python service?", "We have Java and .NET in the same company."
So the community is now AI for Developers. Every skill now detects your stack first and works with .NET, Node/TypeScript, Python, Java/Kotlin, and Go. If you also touch the frontend, or your team uses more than one language, it's all in one place now.
If you're here for .NET: nothing is taken away. .NET is still where I come from and where most of my examples are. There's a dedicated .NET Pack, and an AI in .NET roadmap is coming.
What you get inside
The Claude Code course is just the newest piece. Here's everything else:
- 49 Claude Code skills + 7 agents for any stack. Two commands and they're in every project: query optimization, test generation, auth, caching, OpenTelemetry, Docker, CI/CD, plus agents for code review, architecture, security, database and observability audits.
- AI Roadmap for Developers - 9 written lessons: from "AI-curious" to shipping AI features (MCP, your first AI feature, embeddings, RAG, agents in production).
- AI Coding Agent Starter Kit - set up any repo for AI agents in 15 minutes: AGENTS.md, rules, hooks, commands, skills and subagents, ready to drop in.
- Build AI Features - runnable projects for semantic search, RAG and your own MCP server.
- Build Challenges - 12 projects from Starter to Expert, a new challenge every month.
- Packs - deep dives like the Legacy Pack (a sample app with 19 planted bugs to practice on) and CRUD vs Event Sourcing.
- Resources - CLAUDE.md generator, templates for .NET, TypeScript and Python, the toolkit dashboard.
- Live Office Hours every two weeks. Bring a question or your repo.
- AI Weekly every Monday: what changed in AI and what it means for you.
- And me answering your questions in the community.
All of that for $19/month. I know. I did the math too.
The price goes up as the community grows. If you join now, you keep $19 for as long as you stay.
Install the toolkit, read Module 1 of the Claude Code course, and run one agent on your own code. Then decide.
See you inside, Stefan
P.S. Already a member? The toolkit moved to a new repo and now works for any stack. Run the 1-minute update from the "Install the toolkit" lesson in Start Here, and you'll get all 49 skills under their new names.