Monday, September 21st, 2026
I'm building a language and standard library for the development lifecycle itself, and an open-core ecosystem on top of it. Every Monday I’ll be running a one-man sprint retrospective, and this is what came out of last week's work.
I don't think I ever explained what the product is (genius marketing, I know). So:
How many software development jobs have you had where the pipeline was slow, hard to maintain, or riddled with loose scripts nobody thought much about?
How much do you love maintaining Dockerfiles and custom images, or copy/pasting the same bash scripts into a new repository knowing they’re just going to drift?
What about watching Slack or GitHub Actions like a hawk and pushing fixes to unblock a release?
Ooh, or maybe migrating your entire pipeline to some new system because of a company mandate? Again?
Anyone?
Yeah, me neither. So I'm rethinking the dev lifecycle from the ground up, for the teams stuck living with pipelines like that. I’m building a language to solve CI/CD.
The language's standard library will cover what a pipeline typically needs: validation, reporting, promotion, deployment, monitoring, ticketing, and escalation. It also covers first time setup of a repository. Already, the setup section of my repository's READMEs is "install the language if you haven't already (brew install, cargo binstall, or curl | sh), then run install to set the repository up for local development."
The goal is to develop something that will work with, simplify, or outright replace all the devops in your stack, depending on how you and your team want to use it. Where the language stands out from everything else is its commitment to adoptability, portability, and dead simple extension from existing scripts. You will not need to rewrite your existing dev scripts unless you want to, and you won’t need to bake some new install into your docker images. In fact, I have two aspirations: reduce the need for custom runner images in general, and make it so that a team with existing CI can adopt it and start seeing the benefits of my operations language inside of a week.
This is where I have to admit that last week I ran into trouble with my own pipeline, which is ironic for a man starting a CI/CD company. It is also a fantastic chance to prove my product’s as I develop it.
On Tuesday, I’d begun to notice that my merge requests were eating up valuable runner minutes and keeping approved merge requests starved, which was slowing down merge and preventing my first public release. In addition, jobs had been proliferating for some time, and longer jobs were slowing down heavily at seemingly random intervals.
So, I decided to run a profile over my pipeline.
I scraped everything my forge had on my language’s repository (67 days, about 2,100 pipelines, 27,000 jobs and 18,700 job logs). Here’s the breakdown: