~paul

~paul

@paulvall.dev

generalist 🇲🇽 | guac enjoyer 🥑 | be kind to each other ✌🏻 | @astro.build @bomb.sh @e18e.dev @studiocms.dev @tauri.app

258 followers149 following703 posts

is this the next step? optimizing ai engineering instead of compile time?

Al Merose@al.merose.com

Waiting for two agents to fuzz and fix your code feels like a new type of long compile step.

built a claude code plugin around typesafe's jev, the new model that returns probabilities instead of prose jev-shadow asks it 5 hazard questions about every command claude proposes and logs the answers next to what claude code's own classifier decided. zero latency, changes nothing

github.comGitHub - dreyfus92/jev-shadow: claude code plugin that measures jev before trusting it: background verdicts, joined with the classifier's real denials, and a reportclaude code plugin that measures jev before trusting it: background verdicts, joined with the classifier's real denials, and a report - dreyfus92/jev-shadow

naur said it in 1985, the real product of programming is the theory that forms in the minds of the people who built it. code is just a lossy artifact of that theory. ai made code cheap. it did nothing for theory. we're shipping artifacts faster than anyone can rebuild what they mean.

🥳

Evil Martians@evilmartians.com

Not all heroes wear capes... at least the @e18e.dev team doesn't. They’re on a mission to make the JS ecosystem lighter and faster. So we’re donating to them as part of our OSS donation program. They've already reduced node_modules size and made Node.js tooling much faster, and we appreciate it!

👌deps

@bomb.sh/tty is the low-level layout engine and input parser that powers our forthcoming tui libraries. think we can get this even smaller, but our baseline is now v0.9.0 at 113 kB (0 deps) compared to v0.6.0 at 830 kB (0 deps) @opentui/core at 21 MB (9 deps) ink at 7.8 MB (36 deps)

alex has been doing an amazing job working on this for the past months 🫡

Alexander Karan@alexanderkaran.bsky.social

The @e18e.dev framework tracker is nearing an official release. We have worked hard to include a wide array of stats on meta-framework performance, including data from HTTP Archive. See below for a sneak peek at the percentage of sites in each framework that meet the benchmark. 🧵

in january 2025 nate invited me to contribute to contribute to @bomb.sh back then clack was doing around 2 million weekly downloads. this week it crossed 19.3 million. almost 20 million. i keep thinking about why.

View full thread

and thank you @43081j.com , who i’ve worked with hand in hand this whole time. his work on release cadence, bug fixes, and shipping features has been crucial to getting clack here.

in january 2025 nate invited me to contribute to contribute to @bomb.sh back then clack was doing around 2 million weekly downloads. this week it crossed 19.3 million. almost 20 million. i keep thinking about why.

View full thread

none of this happens without open source. people reviewing prs on weekends. contributors fixing spinners and shipping accessibility work because it matters, not because anyone paid them to.

in january 2025 nate invited me to contribute to contribute to @bomb.sh back then clack was doing around 2 million weekly downloads. this week it crossed 19.3 million. almost 20 million. i keep thinking about why.

View full thread

software is becoming agentic, and agentic software lives in the terminal. every ai coding agent, every scaffolding flow, every automated workflow runs through cli tooling.

the terminal stopped being a niche developer surface and became the primary interface where humans and agents meet. libraries like clack are quietly load-bearing for that entire layer.

in january 2025 nate invited me to contribute to contribute to @bomb.sh back then clack was doing around 2 million weekly downloads. this week it crossed 19.3 million. almost 20 million. i keep thinking about why.

part of it is the obvious stuff. clack makes cli prompts feel good. frameworks and scaffolding tools adopted it. the community kept it healthy. but there’s a bigger shift underneath.

i don’t get GrokBot’s hype aside from cloud agents having their own VMs. the training data grok uses is heavily influenced by elon & garbage culture on X

the npm deprecation banner for upper-case-first is just the replacement code: "Use input.charAt(0).toUpperCase() + input.slice(1)" the whole package was that one line. reviewing replacement issues for module replacements and this one reviewed itself 😅