How to use Claude to revive a Flash game engine
Pulado’s maker used Claude Code to repair a Flash client for Ruffle and port it to Haxe. Repeated feature comparisons and playtests brought the engine close to parity, with remaining issues still under repair.
news.ycombinator.com
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Repair [legacy Flash client] for [Ruffle environment], then port its ActionScript engine to Haxe. Use [existing executable and source] as the behavioral reference. Test each feature, compare both versions repeatedly and playtest [game scenarios]. Report remaining parity gaps before calling the port complete.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- fmxexpress
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- The revived engine was close to ActionScript/Haxe parity, though the author was still ironing out engine issues and generated artwork orientation.
- Human involvement
- The author required repeated feature tests, parity checks and playtests and was still resolving engine and artwork issues.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Had Claude Code fix the Flash client to work in Ruffle.
- Had Claude Code convert the engine to Haxe.
- Repeatedly tested features, compared ActionScript and Haxe behavior and playtested the engine.
Pulado was an older Flash arcade game engine that had become unplayable after Flash support ended. Its dynamically loaded remote assets prevented Ruffle from running it successfully. The author reports using Claude Code first to fix the Flash client for Ruffle, then to convert the engine code to Haxe as a second version.
The conversion needed repeated correction because Claude made wrong assumptions about how the engine worked. The author had it test every feature, compare the ActionScript and Haxe versions repeatedly and playtest the engine. The reported state was close to parity, with remaining engine issues still being resolved. The author describes the project as alive again rather than presenting the port as finished.
On the editor side, the author also reports a rewrite from ASP into JavaScript. An AI agent helps configure settings and generate fresh artwork alongside the manual editor. Artwork remained another unfinished area: the engine imposes orientation rules that pixel-art generation models struggled to satisfy. The post names Claude Code for the client work but does not identify the editor's assisting agent. Who it's for: maintainers reviving legacy game tools who can compare behavior against an existing implementation and conduct repeated playtests.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Hacker News · fmxexpress
Frequently asked questions
Why did the Flash client need repair?
The author says dynamically loaded remote assets prevented Ruffle from running it. Claude Code fixed the client for Ruffle before the Haxe conversion.
Was the Haxe port finished?
The author described it as pretty close to parity and still having engine kinks. Repeated feature testing and playtesting were needed because Claude made wrong assumptions.
More details
Remaining steps
- Rewrote the ASP game editor into JavaScript.
About this account
Claude Code repaired and ported the client, but incorrect assumptions required repeated direction and feature-level testing.
Platform: Hacker News. Attribution: https://news.ycombinator.com/item?id=49691938. Prompt: Suggested (written by Agents In Use, not the author's). Sourced by the Agents In Use pipeline.
Credit: fmxexpress on Hacker News
Source published
Added to Agents In Use
Source checked
Tools and topics
Claude Code · Ruffle · ActionScript · Haxe · JavaScript
Flash migration · Haxe · legacy games
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.