voidmain42

I'm pretty late to this thread, but I battled with the whole "keep them organized" issue across 20+ repositories. I see a lot of folks suggesting scripts and symlinks, but I stopped using that approach because filtering (i.e. figuring out which files belong in each repository) and syncing became a headache.

The Agency HQ (https://theagencyhq.dev) is another approach that uses a central repository of skills, rules, and other files that are pushed down to developer machines and then placed into project directories as needed. It uses a simple daemon on to pull down the files and distribute them. And it supports filtering and translation for most of the popular agents.

Disclaimer, I wrote The Agency and it is in early stages. I'd love feedback if anyone gives it a test drive.

Hi HN. After exiting FusionAuth, I’ve been building a lot with coding agents, and the config sprawl got out of hand.

I’d fix something in one repo and forget the other ten. Or I’d start a new repo and have to copy everything over from whichever repo had the latest version.

I tried symlinks, submodules, subtrees, copy scripts, hooks, launchers, and a few other things. Nothing really worked.

The Agency lets you define config once and sync it to every repo that needs it. A small background daemon runs on every developer’s machine. It handles retrieving the latest config and distributing it to every repo that’s cloned on their machine. It removes drift issues entirely. Here’s a diagram that explains the flow:

https://theagencyhq.dev/how-it-works/

It’s day one, so it’s rough. It runs on Mac and Linux and Mac is the best-tested platform. I’d genuinely like to hear where it breaks for you.

Yep. Westminster technically.

That's the company I founded. I wrote most of the tools, frameworks, and libraries it uses.

It's an ecosystem that includes a number of tools, libraries, etc. The organization (include the GitHub org) is called Latte. Each project is a GH repository under that.

The dependency management system is part of the `cli` tool. And the repository management is the `app` project and published to https://app.lattejava.org.

The `cli` tool is the foundation though. It's how projects start, build, run, release, and publish.

Hope that helps clarify things.

I find that Maven and Gradle are quite complex. Creating a build file for a new project and publishing to Sonatype is extremely difficult.

Certainly those projects are solid, but I think there is a lot of room for improvement, and that's what I'm working on.

I think that's why other platforms have done so much better than Java. NPM, Python, Ruby, DotNet, etc.

They all have dependency management, versioning, publishing, repository management, HTTP, web frameworks, and build tools as one complete ecosystem.

Latte's core systems are also battle tested (http, cli, jwt). They are forked from projects I built for FusionAuth, which supports around a billion users and has millions of downloads and tens of thousands of production deployments.

I'm the author of Latte.

Some of the projects are actually forks from some of my open source from the company I founded (and later sold). Some have been around for quite a while (10+ years) and have some adoption as well.

To keep those projects going, I forked them and started Latte.

My plan is to keep working on all the projects and really start promoting it. I think it's a game changer for Java.

I'm currently working on adding HTTP/2 to our HTTP server. It's nearly done and looking really solid. And our benchmarks on it are looking like we'll be the second fastest HTTP server for Java.