Founder @ JS Mastery @GitHub Star, software engineer & educator Agentic Engineering Course is live now, 20% off through Sep 29

Croatia
New YouTube video just went LIVE! 🔥 What if you could go from typing "build this feature" and hitting enter to actually knowing how to work with an AI coding agent? 🧵👇
2
2
23
1,407
When an AI coding agent is about to make a big change, like adding a database, have it show you a plan first. The cheapest place to fix a wrong decision is the plan, not the code.
7
1
20
1,142
Just one hour to go! 🔥 Here are some final FAQs if you're still on the fence: 1) “I use Cursor, not Claude Code. Is this still for me?” Yes. Absolutely. We cover the broader AI coding stack throughout the course. Claude Code just gets a deeper treatment because, right now, it gives us the most control over how agents actually work. But the important part isn't the tool. Spec-driven development, structured review, keeping a record of decisions, knowing when to trust the agent and when to stop it- those are skills that carry across tools. So if you use Cursor, Codex, Claude Code, or something else six months from now, the workflow still applies. The tools will keep changing. The way you work with them is the part that matters. 2) “How long does it take?” There isn't a correct answer. Some people in the early group went through all 8 modules in under two weeks. Others are taking it much more slowly. Both are fine. You get lifetime access, and the course will keep getting updated as the tools change. 3) “I work full-time. Is this actually realistic?” That’s how most of the early group did it. The modules are designed so you can disappear for a week, come back, and still know where you left off, and I would say the best way to learn is to apply it while you're working. 4) “Should I get the course or the bundle?” The course now gets you all 13 modules, the skills repo, and lifetime access. The bundle adds the live workshop, where I take a real codebase and run the entire workflow live, including the decisions, mistakes, trade-offs, and questions along the way. The workshop has limited seats and only exists during this launch. If budget is the deciding factor, get the course. If you want to watch the whole thing happen on a real codebase, and ask your questions live, get the bundle. Basically: course = learn the system. Workshop = watch me stress-test it in public. 5) “I’ve already taken AI courses. Convince me this is different.” I won't. Here’s the difference instead: Most AI courses teach you what the tools can do. This one is about what happens when you start using them to do real work. How do you structure the work? How do you review what the agent produces? How do you know when it’s confidently wrong? How do you keep decisions from disappearing into a chat history? How do you build a workflow that gets better instead of starting from zero every time? That’s the gap we built this for. 6) “What if I get in and realize it’s not for me?” 7 days. Full refund. ----------- If you’ve been going back and forth, you probably have enough information by now. The rest is just deciding. Join now at 20% off 👉 jsm.dev/ai-course-twitter
2
9
1,313
Anyone can prompt Claude Code and get an impressive result in 30 minutes - completely missing what's actually holding it up underneath. The 85% below is everything that decides whether it survives contact with real users: ✅ What to build ✅ How to structure it ✅ What happens when it fails ✅ What happens when it grows ✅ What happens when another developer opens it in six months(!) "Write better prompts" doesn't touch the 85%. This course is the 85%. Just 2 hours to go until the launch discount and workshop are GONE! 👉 jsm.dev/ai-course-twitter
1
2
10
1,281
AI education is broken. There's an ocean of content on using AI to code. Almost none of it connects into something you can actually run on a real project, from the first commit to the thousandth. You end up with a pile of tips and no workflow... The Ultimate AI Engineering Course is different. Grab 20% off here 👉 jsm.dev/ai-course-twitter
Made with AI
4
1,112
When you tell an AI coding agent how to design your UI, vague adjectives don't work. "Make the landing page premium" tells it nothing. "Use large product imagery, generous spacing, and restrained controls on storefront pages" is something it can actually follow.
6
24
1,683
Last chance to grab the Ultimate AI Engineering Course at the launch discount (20% off) - and the final day to sign up for the LIVE workshop ➡️ jsm.dev/ai-course-twitter
1
1
958
Job security speedrun: learn AI engineering ✅ → jsm.dev/ai-course-twitter
1
3
1,212
"Senior Staff Backend Engineer" doesn't make Claude any smarter. A subagent isn't a genius employee. It's a separate context. Use one when work can happen on the side, so your main conversation doesn't have to carry it.
1
10
1,376
It turns you from someone who prompts and hopes - into an engineer who scopes, builds, reviews, and ships with a repeatable system → jsm.dev/ai-course-twitter
4
1,178
Pretty cool how Opus 5.5 created this promo for our agentic engineering workflow in a single prompt!
2
1
17
1,566
Start a fresh project, describe what you want, and the agent scaffolds a working app in minutes. Then you point the same agent at the codebase you actually get paid to work on. And the magic is gone. 🧵👇
4
3
17
1,761
7/ The fix: make the codebase legible before the agent touches it. Write a map for the agent: the real module boundaries, the patterns your code actually uses, the landmines, how data really moves. Keep it in a file the agent reads on every task.
1
124
8/ Then pin down current behavior with tests before you change anything, and give the agent more room only as the system gets more legible. A fresh project has no history to get wrong. Yours has ten years of it. Teach the agent your history, or watch it confidently guess. Read about this in more detail here ➡️ jsmastery.com/blogs/the-ai-t…
123
As AI writes more code, the developers who direct it and catch its mistakes are the ones who stay valuable. ➡️ jsm.dev/ai-course-twitter
1
1
10
1,381