A small app that ships with AI agent skills to extend and audit it

Hey HN long time lurker, first time poster.

I built a small web app for practicing the rough estimation math that comes up in system design.

The app itself is really simple: one HTML page, no framework, no backend, and no build step. It gives you 10 random prompts per round, accepts simple expressions as answers, and shows reference answers and explanations.

But the main thing I wanted to experiment with is not the app itself.

The repo also includes bundled AI agent skills that can:

- Use WebSearch to add new questions for a topic

- audit the question bank for math, clarity, and fairness and optionally fix issues

So instead of shipping just the source code or a binary, I'm shipping the ability to maintain and extend the project as part of the project as first-class artifacts that live alongside the code.

In this case the app is for system design estimation practice, but I think the pattern could apply more broadly to a lot more use cases and projects.

Curious what people think about both:

- the app itself

- the broader idea of shipping a project together with agent-operable skills for extensibility and maintenance