Lachlan Donald

@thdxr Agreed, it's agonisingly slow. Was just ranting about it this morning: x.com/lox/status/203…

i still don't get why we need to push code up to get an LLM review via awkward github ui hacks opencode has /review which can also do things like run your code to check things but a full time team focused on this would do it better, i just don't like the workflow they offer

PRs are too slow for agentic dev, but we still need a way to control which commits get CI credentials. Today’s loop is awful: push, wait for review, poll, fix, push again. I think review + CI should produce signed attestations attached directly to commits, so trust can be established before push and CI doesn’t have to wait for a post-push approval loop. Exploring this in github.com/buildkite/git-…. Anyone interested?

@ryanvogel We exist in the future too.

everyone is trying to build async agents that work when they sleep but all they really need are Australians

Now paying Granola $18/mo to upload my meeting audio to their servers. So I built my own thing. It's 7 MB. Everything runs locally. Works with @tobi's QMD for super fast search, PARA method second brain setup for organized people, projects, entities, @obisidan, CLI, claude code, codex, Claude Desktop. Before a call with Biff it pulls your last 12 conversations, shows you pricing came up 5 times in 2 weeks, and reminds you that you owe Biff a doc from Friday. After the call it asks if you got what you came for. You just ask "what did Biff say about pricing last week" and it pulls from your transcripts. No API key needed, use your existing subs. The meeting prep feature came from messing around with @garrytan's gstack skills. His /office-hours pattern forces you to be specific instead of accepting vague answers. I stole that idea and applied it to meeting prep and post-meeting debrief. It's called Minutes. Rust, whisper.cpp, Tauri. MIT licensed. First real release on github.

Pretty tempted by @pierrecomputer, any interest with making the integration with @buildkite even better?

I am so sick of @github. If I have to see another outage or a rate limit error where it's not possible to increase the rate limit, I am going to build my own. This is absolutely not an adequate platform for agents to ship code fast, let alone humans. We need better.

@jamonholmgren The answer is to hire Australians, we both work in the future and the night.

Tech companies pay a lot of money for CI servers and then have a bunch of super powerful Mac hardware sitting idle for 16 hours a night. Hm.

Have been really enjoying using mise en place at @buildkite recently (thanks @jdxcode). Figured I'd better add a :mise: emoji: github.com/buildkite/emoj…

@mitchellh @jfroma I’m working on github.com/buildkite/clea… which is at least spiritually inspired by vagrant and friends

@jfroma Vagrant itself is in the IBM machine now but the idea of Vagrant adapted to contemporary patterns would absolutely explode right now. I've talked to a LOT of devs (and VCs lmao -- not interested on my side) about it.

I think is time to bring back Vagrant cc @mitchellh

great 2 weeks to be an Australian if you use Claude Code 😂

@grhmc @buildkite Let’s see if we can help!

@lox @buildkite btw I’ve been thinking about bringing this workload back “home” to buildkite recently, so … 🥺

CI services should allow the task to emit multiple streams of output to solve the interleaved text problem.

@thdxr I assume for always hot caches?

who is gonna bring back stateful CI i'm ready to spin up jenkins again

your biggest enemy is still complexity. it's also your agent's biggest enemy. but it has no holistic view of your code base, so it keeps adding complexity. and you think that's how it's supposed to be, because the clanker shat it out, and you don't know the stack. glhf!

@gregpr07 this may surprise you that thus is coming from me but I think we’re in for a 1-3 year period where stuff might break at 3am and if you’re relying on loops to fix it and nobody understands what’s under the hood, you’re looking at an existential threat to your company

Wrote a blog post about my journey here. Has some scalability limitations & will fix them soon. Appreciate any pointers/feedback! jytan.io/blog/firecrack…

The people I want to hear from right now are the security teams at large companies who have to try and keep systems secure when dozens of teams of engineers of varying levels of experience are constantly shipping new features

we've never done code review but damn if your team is producing this much code you're using LLMs entirely incorrectly no one struggles with large amounts of code more than an LLM, if you don't keep that in check you have a self defeating codebase

Lately I’ve been going deep on agent harnesses and sandboxing (turns out there is a lot of overlap with CI). I kept rebuilding the same loop plumbing, so I pulled it into a small Go library: github.com/lox/agent-harn… Drew a lot of inspiration from @Hxlfed14's post 👌

It's still broken, what is happening folks?

@hellostake your app is broken, your website is broken, what is going on? Seeing lots of 500 errors from backend, so it's not a browser or client issue as far as I can tell. Do you have a status page?