Pretty sure this menu is burned into my screen.

TIL typescript checker is just one massive TypeScript file. ~54k lines of code, jfc.

.Raycast focus has been the best pomodoro/focus app I've used. Easy to start and add/remove blocked apps. Just makes it easier to focus and track what I'm doing.

How did I not know about ctrl + r in the terminal for searching terminal history. No more ⬆️⬆️⬆️

While I sit watching cursor AI implement features I feel both extreme excitement for the future of possibilities and a deep existential dread for the things I used to enjoy being automated away.

I can't really understate how much a rollercoaster of emotions I've been on with AI lately. Literally everyday it swings from "holy shit this is amazing" to "we're so screwed".

This. I've been feeling this way for a while now and trying to work through it and find some pleasure in prompting and reviewing/tweaking code. Unsuccessful so far but will keep at it because this is the future.

One skill I'm developing because of cursor agent is reviewing code. So much code to review and always some weird finds like static variables to get around certain checks.

Something I've been noticing lately. The more coding experience you have, the less you enjoy vibe coding. The opposite is also true. Do you agree?

For every successful dropout there are probably 100s of others that regret it.

The easier you make it to write code the less incentive people have to go out and learn the fundamentals. Sneak peak at a upcoming episode where Matt Lawrence ☕ and I discuss the impact of AI on developer education.

For serious AI and MCP development the standard seems to be Python. Is JavaScript just not favored by those devs or is there a specific reason why Python is a better language for this?

Steps I take before starting to work on a feature 1. Pull latest from dev branch 2. Create new feature branch and switch to it 3. Spin up the dev server to make sure everything is working at baseline 4. Document changes I'm going to make in a readme (probably deleted/refactored

This right here. All it takes is just trying to use it in a project that has actual users and a slightly larger scope and you'll quickly be wrangling the AI for more time then you'd spend writing some features. To properly use AI you NEED to know it's limitations more so then

Everything hinges on how fast and well AI progresses. Right now it's a tool for us to use. Agents are cool but they aren't replacing anything. If they get even 50% better though...

The hardest part of engineering leadership is adding in processes and guardrails that make the codebase more maintainable but also let your team ship. I definitely lean on the side of less process to allow for higher velocity but as the codebase matures you need to add more