Ask HN: What Are You Working On? (March 2026)

working on an AI-native Kubernetes sidekick that watches your pods, reads the logs, and turns failures into clear fixes before they become outages

Each game adds more building blocks to the editor: multiplayer, event systems, NPC behaviors, pathfinding, etc. I build a system once, and then anyone using the editor can use it in a click.

Since my last month, I shipped the asset marketplace and the LLM builder. Artists can now upload tilesets and characters, and unlike itch.io, assets drop directly into the editor. You can preview how they'll actually look in-game before using them [1].

An other problem I kept running into: even with a no-code editor, users don't know where to start. So now I'm extending it with a coding agent. Describe the game you want, and it assembles it — pulling assets from the marketplace, wiring up the event system, and using all the building blocks I've spent the past year extracting. Multiplayer, mobile controls, pathfinding, NPC behaviors — the agent doesn't build any of it, just reaches for what's already there.

Once the LLM assembles it, users will have a game ready to work on, and will still be able jump into the editor and tweak everything [2]. Here's an example of what it can already make [3] (after a lot of prompting), and the goal is to reach games like this one I built with the manual editor[4].

Hoping to release the AI mode in a week or two. The manual editor is live at https://craftmygame.com in the meantime.

[1] https://craftmygame.com/asset/mossy-cavern-JdYWai1

[2] https://youtu.be/6I0-eTmoHwQ

[3] https://youtu.be/FZ12XSZu4nM

[4] https://craftmygame.com/game/island-survivor-s1Ay7Go

Today, if you search for "what size should I get in Nike Air Max 90" you'll find size charts. We have it, and for 200+ brands across 70+ retailers. When users tell us which shoes they own and what size fits them we’re slowly building crowdsourced fit recommendations which are personal and more accurate compared to size charts.

We're two coders who've built an almost fully autonomous platform. AI agents build, debug and deploy crawlers on their own. We went from 4 crawlers to 280+ in about a month, and the whole thing runs on a home server. When new shoes are discovered, the platform publishes new pages with relevant info automatically. Agents get access to platform metrics and SEO data via custom MCPs to identify the right opportunities on their own. Currently at about 3000 MAU and about 100 size recommendations/day.

Example: https://www.getsize.shoes/en/shoes/nike-air-jordan-1-low-se-...

A lot of existing databases are storage first, with everything else built around them. I have been exploring what it looks like if the database is closer to the application runtime itself, where state is live, queryable, and easier to reason about directly.

One thing I am prototyping right now is database-native tests.

Basically: what if integration tests were a database primitive?

CREATE TEST test::insert { INSERT test::users [{ id: 99, name: "Ghost" }]; FROM test::users | FILTER id == 99 | ASSERT { name == "Ghost" }; };

So not a wrapper, not a framework, not an external test runner.

A real test object inside the database.

The idea is that you could run these before schema changes, and make stored procedures or other database logic much easier to test without leaving the database model.

Still early, but it feels like one of those things that should just exist, especially for databases built around live application state.