Agents In Use
What’s new todayLast update
108 use cases added today

How to use Claude to build cross-platform software

The author built a multi-target compiler, used it with Claude for a WASM client and server sharing classes and data formats, and began a cross-platform UI framework.

Source published news.ycombinator.comAgent: Claude
OUR ADAPTATION

Suggested prompt

Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.

Help build [cross-platform language toolchain] for [host platforms] and [target platforms]. Use it to implement [WASM client] and [server backend] with shared classes and data formats. Then develop [UI framework] around shared program logic and platform-specific views, following my design and checking each result.

Evidence at a glance

Sourced account
Source author
spacedcowboy
Original account
Read the source ↗
Published
Source link checked
Reported result
The author reports a signed phone binary developed on Linux, shared client/server classes and data formats, and an ongoing cross-platform UI framework project.
Human involvement
The author says they make sure the result works, but does not describe specific review procedures or test suites.

A linked account supports attribution. It does not mean we independently reproduced its result.

Report an issue with this caseOpen correction email

Opens your email app with the case link included. Nothing is sent automatically.

How the agent did it

  1. Built a compiler for their own language across Mac, Linux and Windows.
  2. Used the compiler with Claude to write a WASM client and server backend sharing classes and data formats.
  3. Began building a cross-platform UI framework with visual connections to shared program logic.

The author describes a development approach in which they supply the ideas, workflow and product vision, hand time-consuming implementation work to an LLM and make sure the result works. Using this approach, they built a compiler for their own language. It runs on Mac, Linux and Windows, with any of those hosts able to produce code for targets including WASM, iOS, Android, Zynq, m68k and 6502.

One reported result is a signed binary running on the author's phone, developed on Linux without an Apple Xcode installation. The compiler then supported work with Claude on a WASM client and server backend. Both used the same classes and data formats, which the author says made integration trivial.

The next project was a cross-platform UI framework, still described as work in progress. Its intended interaction uses a Cocoa-like UI description and visual connections between shared program logic and desktop, tablet or phone interfaces, including portrait and landscape mobile variants. The author says they would not have considered a program of that complexity without the LLM. Who it's for: experienced developers pursuing personal language and platform tooling while retaining responsibility for design and checking the result.

Prompt
Suggested prompt (written by Agents In Use, not the author's)

Source: Hacker News · spacedcowboy

Frequently asked questions

How were the client and server related?

The author says they and Claude wrote a WASM client and server backend using the exact same classes and data formats, making integration trivial.

Was the UI framework complete?

No. The author describes working on it, with shared program logic visually connected to desktop, tablet and phone UI variants.

More details

About this account

The author handed implementation work to the LLM while retaining ideas, workflow, vision and responsibility for the end product.

Platform: Hacker News. Attribution: https://news.ycombinator.com/item?id=49672344. Prompt: Suggested (written by Agents In Use, not the author's). Sourced by the Agents In Use pipeline.

Credit: spacedcowboy on Hacker News

Source published

Added to Agents In Use

Source checked

Tools and topics

Claude · custom compiler · WASM

cross-platform · compilers · WASM

Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.