Top AI Product

We track trending AI tools across Product Hunt, Hacker News, GitHub, and more  — then write honest, opinionated takes on the ones that actually matter. No press releases, no sponsored content. Just real picks, published daily.  Subscribe to stay ahead without drowning in hype.


Xmloxide: The XML Library That an AI Wrote From Scratch — and It Actually Works

There’s something almost absurd about the story behind [Xmloxide](https://github.com/jonwiggins/xmloxide). A developer pointed Claude Code at libxml2’s test suite and basically said “make this pass.” Three hours later, the agent had produced a full Rust reimplementation of one of the most important XML libraries ever written. And here’s the kicker — it passes every single test. All 119 in the libxml2 compatibility suite. All 1,727 in the W3C XML Conformance Test Suite. A perfect score.

The project [hit Hacker News today](https://news.ycombinator.com/item?id=47201816) as a Show HN post, racking up 58 points and 60 comments worth of lively debate. Some folks were genuinely impressed, others raised valid concerns about whether AI-generated code should power production systems without heavy disclaimers. The original libxml2 maintainer even showed up in the thread to clarify that SUSE has picked up maintenance, though the memory safety issues that plague C code remain.

And that’s really where xmloxide gets interesting. libxml2 became officially unmaintained in December 2025, and while it’s been patched over the years, it still carries the inherent risks of a massive C codebase. Xmloxide sidesteps all of that with an arena-based architecture, zero unsafe code in its public API, and no global state. Every document is self-contained, thread-safe, and Send + Sync out of the box. No more calling initialization or cleanup functions.

The feature set is surprisingly complete for something generated in a single session. You get DOM tree parsing, SAX2 streaming, a pull-based XmlReader, XPath 1.0 evaluation, HTML 4.01 parsing with error recovery, DTD/RelaxNG/XSD validation, and even an `xmllint`-compatible CLI tool. Performance benchmarks show serialization running 1.5-2.4x faster than libxml2 and XPath queries clocking in at 1.1-2.7x faster.

Is it production-ready? The creator is honest about this being an experiment, not a decade-long commitment. But as a proof of concept for what AI agents can do when given a clear target and a robust test suite, xmloxide is hard to ignore. Whether you care about XML parsing or not, this one’s worth watching.


Discover more from Top AI Product

Subscribe to get the latest posts sent to your email.



Leave a comment