Michael Dwan

I feel this. As soon as I hit the character limit I say fuck it and go do something else instead.

The Claude Code extension for VS Code/Cursor is an acceptable middle ground. Prompts to kick it off, diffs to follow along & steer it. Still a long ways off from a "good" dx though.

API errors from Claude

Me: I'm finally done with meetings, time to code!
Claude Code: Whelp you're driving. Hope you still know how.

The Spiderman pointing meme, but with Pac-Mans

I landed somewhere similar after testing AI code review bots. I wish giving humans super powers was as lucrative as trying to replace them.
bsky.app/profile/mich...

This is so cute. But I get too many newsletters already. But I’m compelled to try it. Grr

Yes! LLMs writing code isn’t very interesting since so little of swe is about writing code. What would be truly uplifting is an AI that helps me understand stuff, especially code other people wrote

What I actually want is for humans to review code with less effort. LLMs can help with this! We don’t want summaries, we want context. A diff shows what changed—help us understand why. Highlight what matters. Explain intent. Code reviews should be a conversation.

Been trying out Copilot PR reviews. After ~10, it found a couple typos and one bug—not bad. But calling that a review is disingenuous. Static analysis + engaged humans nailed this ages ago, it was just time-consuming. I wish AI aimed to make good things stupidly easy—not replace them with worse.

Dockerfile Explorer | Depot A tool for introspecting the LLB output of BuildKit's Dockerfile parser.

Ever wondered how BuildKit—the engine behind `docker build`—turns a Dockerfile into an image? 

depot.dev made an interactive explorer showing how each line of a Dockerfile (the frontend) maps to a low-level-build operation in BuildKit (the backend): depot.dev/dockerfile-explorer

yes!

Thanks I’m giving it a try now!

My typical ChatGPT workflow:
1. start new chat, pick a model
2. model names make no sense, stick with the default
3. ask question, get underwhelming answer
4. try claude, cursor, or copilot and move on
5. see people on here raving about ChatGPT
6. GOTO 1

PSA: remember it's important to lock github actions to an immutable commit hash so that the node.js script that fetches and executes a shell script from a random attacker-controlled URL cannot change 💪

I’m excited to ship tools that give AI engineers more power while also being "boring" enough that the rest of us can actually use it to build things that matter.

Can't wait for you to see what's next!

I read recently that of the ~30m software devs around the world, only ~500k are "AI engineers"—that stuck with me because "AI" is something we all need to harness—for our products and our productivity. But it's too damn hard!

What sealed the deal for me: Replicate has a high bar for software quality and developer experience. They sweat the details—and I love that. Good taste matters in a crowded market!

Exciting news: I’ve joined @replicate.com to work on dev tools for AI!

I’m thrilled. Replicate is taking off in a wild space full of opportunity. The platform is excellent and the team is sharp, kind, and thoughtful about how they build both product and company. It’s a joy to be in the mix. 🧵↴

Barely... but you can uninstall system apps as of a few years ago. Removing safari, mail, etc goes a long way

The Sim City disasters menu with options for Fire, Flood, Monster, Riots, and more.

we've reached the point in our simulation where a bored player triggers disasters just to see what happens

old iPhones locked down with screen time work pretty well! Not sure about Spotify, but Apple Music can airplay to sonos and homepods on ~iOS 12

Beautiful

Dog Man graphic novel addresses issues with self confidence. "We're all in jails of some kind... imprisoned by the stories we tell ourselves: I'm not good enough... I'm not smart enough... I've made too many mistakes. But Sprinkles... Freedom is in our minds, too."

This Dog Man graphic novel gets it

I'd read the shit out of your food blog

Another use case that I think has traction is WASM as a portable sandbox for untrusted code. If you can target WASM you can run on lambda, CF workers, embedded scripting, etc. Lots of applications but still niche for something that was supposed to be the second coming

It’s harder to retcon than bury history and start new. Esp when the department of education is a propaganda arm. Crappy analogy: my kids don’t know a world where Han Solo shot first because all that’s available are the altered films.

I love reading your retrospectives. Thanks for sharing!

Strict tables should help, but it’s off by default and requires a change to CREATE TABLE. :old-man-shakes-fist-at-flexible-types:

runtime: a Windows application launched via Steam sometimes freezes · Issue #71242 · golang/go Go version go version go1.23.2 windows/amd64 Output of go env in your module/workspace: set GO111MODULE= set GOARCH=amd64 set GOBIN= set GOCACHE=C:\Users\hajimehoshi\AppData\Local\go-build set GOEN...

This “Steam overlay breaks the Go runtime” bug is wild. 1) it’s concerning that Steam messes with low-level APIs like this, and 2) I love reading about this kind of debugging nightmare as an unimpacted bystander 🫠 Glad the Go team is on top of things like this github.com/golang/go/is...

Sounds like they hate their veggies