RT @NickADobos: Grok 4 announcement recap. I watched 1 hour of an awkward rambling demo so you don’t have to!. - 2 new models, grok 4 and g….

Spotted deep in the vibe tunnel 👀

WOW.

RT @VibeTunnel: If you wanna contribute, there's a lot of issues that would work well for new contributors: 💚

RT @VibeTunnel: New vibe bits are out! 1.0.0b7! ✨

X blows up when Apple reveals Control Centre with Liquid Glass, X commentary points out poor contrast, Apple fixes it, X blows up because it doesn’t look like glass anymore.

Sad vibes on the train 🚊

Had an awesome time hanging out with @steipete today.

The MCP vs CLI debate is heating up 🔥, and as promised I said I’d share my view. Well, here it is, enjoy! 🙂.

They're not, well I don't think so anyway:

My Take on the MCP vs CLI Debate.Personal thoughts on the growing CLI movement and why I still see value in MCPs for certain use cases.

Claude Code Top Tip: Double Tap Esc to jump back to a previous message and fork the conversation. Great for when you go off on a tangent and want to get back on track.

"Based on median token usage, the Pro plan currently covers about 225 Sonnet 4 requests". So it's more cost-effective to switch back to request-based pricing if using Claude Sonnet, as you'll roughly receive half the amount of requests under the new pricing mode?l 🤔.

So you still need an Observable model then whether that’s called “SomeModel” or “SomeViewModel” in order to encapsulate the network requests and to hold state that is associated with the result of that request in order to test. So either we don’t test or we just create a view.

I can highly recommend SnapAI, by @betomoedano, which solves this by using the new `gpt-image-1` model, which avoids the need to remove the background

"I’m not saying all MCPs are useless - for example Microsoft’s Playwright MCP for browser automation is great. However, they also built an MCP for GitHub, which is simply a lesser version of the existing gh cli which does the same thing.". @steipete, you mentioned Playwright but.

I’m so glad to hear that. That’s exactly why I built it.

“Claude knows how to compile Swift packages via swift build, but for my macOS application target, it often failed to figure out the right xcodebuild invocation. XcodeBuildMCP solves the problem by giving the model a simplified set of tools for building and running apps.”.

So MVVM skeptics, how do you unit test logic and state on the view? Say you have a fetchData() function on the view that updates @State var someProperty and using .task modifier to kick off the operation. SwiftUI docs says you’re not allowed to access or mutate state outside of.