Show HN: Sgai – Goal-driven multi-agent software dev (GOAL.md → working code) (github.com)

36 points by sandgardenhq ↗ HN
Hey HN,

We built Sgai to experiment with a different model of AI-assisted development.

Instead of prompting step-by-step, you define an outcome in GOAL.md (what should be built, not how), and Sgai runs a coordinated set of AI agents to execute it.

- It decomposes the goal into a DAG of roles (developer → reviewer → safety analyst, etc.) - It asks clarifying questions when needed - It writes code, runs tests, and iterates - Completion gates (e.g. make test) determine when it's actually done

Everything runs locally in your repo. There’s a web dashboard showing real-time execution of the agent graph. Nothing auto-pushes to GitHub.

We’ve used it internally for prototyping small apps and internal tooling. It’s still early and rough in places, but functional enough to share.

Demo (4 min): https://youtu.be/NYmjhwLUg8Q GitHub: https://github.com/sandgardenhq/sgai

Open source (Go). Works with Anthropic, OpenAI, or local models via opencode.

Curious what people think about DAG-based multi-agent workflows for coding. Has anyone here experimented with similar approaches?

10 comments

[ 202 ms ] story [ 1268 ms ] thread
author here!

I have been working on this for my own use until recently, when I shared with the rest of the team, and we thought it would be nice to let the world see it.

I have been interested in autonomous code development for quite some time (at least since March/April 2025) - and summer '25 is when I felt the models were good enough to be pushed to autonomy.

I wrote a bit about it[0], and sgai is the incarnation of my take on AI autonomous coding.

sgai is not even v0 yet, a lot of work to be done to improve its implementation - but I think it should be usable enough for those willing to give it a try.

0: https://cirello.org/aifactory.html

Interesting license choice, modified MIT it seems, with this additional clause:

> No licensee or downstream recipient may use the Software (including any modified or derivative versions) to directly compete with the original Licensor by offering it to third parties as a hosted, managed, or Software-as-a-Service (SaaS) product or cloud service where the primary value of the service is the functionality of the Software itself.

Doesn't that kind of conflict with the "including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software" part of regular MIT, which your custom license also includes for some reason?

I think you might be better of with just not trying to do it "kind of open source but also not" and just say "Copyright 2026 Sandgarden.com" or whatever, instead of the mix of proprietary and open source. Then you get 100% "full control" over what people can do with the source, and don't have to worry about anything when it comes to licensing :)

Sgai (pronounced "Sky") - That's a bit of a stretch
Very nice, the automated setup instructions for opencode are a genius touch, more people should do that.

Is this already your daily driver for coding projects?

Pretty cool

Your goal.md examples are all features for the existing codebase. Any largish goal.md examples where your system is able to 1 shot a pretty large app?

The goal.md is what makes this thing either amazing or terrible for the user, so any guidelines or clear examples on writing a good one would go a long way.

wow bro, very smart of you, can you please share the apps it built, I am very curious to see how it performed. will try it out when I can get some api credits...
the DAG decomposition approach is interesting — curious how it handles goals that span multiple services/repos. i build a multi-service platform solo with claude code and the hardest part isn't the coding, it's knowing which files across which repos need to change for a given goal. do you see sgai supporting multi-repo goals, or is it scoped to single-repo for now?
Sgai pronounced “Sky.” The letters “S” and “G” just showed up to watch.