Cloudflare's tsrs: A Guide to Evaluating the TypeScript to Rust Port

Cloudflare is behind tsrs, a TypeScript to Rust port made by agents, for agents. Here's how to carefully evaluate it before touching your project.

AI for developers

Cloudflare has made a move with tsrs, a TypeScript to Rust port that, according to The New Stack's report, is written by agents and designed so that other code agents can verify its work without devouring computing resources. The article's title already clarifies one thing: it's not a TypeScript fork.

We won't invent usage instructions that the material doesn't provide. What we can give you is a reproducible method to follow the project, read the primary sources, and critically decide if you're interested in trying it. This is an evaluation guide, not an installation tutorial.

What you need

All you need is your own TypeScript project where you can experiment without risk, a separate branch, and access to the sources. Keep the tsrs repository, the Cloudflare blog post, and the comment threads on Hacker News, where the community often points out gaps, handy.

A bit of historical context is also helpful: TypeScript arrived in 2012, and its official line is progressing with TypeScript 7.0. Knowing what the official version does serves as a benchmark for any alternative.

1. Read Cloudflare's blog post and note what problem they claim to solve and what they promise.
2. Review the tsrs repository: license, status, documentation, and how they describe its limitations.
3. Consult the Hacker News thread and its notes to see criticisms and doubts from other developers.
4. Compare it with related projects like Oxc, whose ecosystem is linked to the repository.
5. If you decide to try it, do so in an isolated branch and compare results with the official compiler before touching your continuous integration.

From TypeScript code to the Rust port, with agents in between.

How to read a claim like this

A port made by agents raises a practical question: how do you know it behaves the same as the original? Look in the repository and blog for what test suite they use and what part of the compiler's behavior it covers. If the material doesn't explain it, treat it as an unknown, not a fact.

This isn't the only case of this pattern. The report itself recalls that Anthropic had already used Claude agents in a Rust port, and that React has also resorted to AI for its own changes. Looking at how those projects managed verification gives you references to demand the same from tsrs.

# Evaluation notes template (not a command)
project: my-ts-app
branch: tsrs-test
questions:
  - what tests support equivalence?
  - what behavioral differences are documented?
  - what about my plugins and my configuration?
result: pending

Tip Before comparing performance, compare results: that the same code produces the same diagnostics and the same output. A faster compiler that makes mistakes without warning is costly. And remember that what's announced isn't universal: what works for Cloudflare might not fit your project.

A simple checklist prevents enthusiasm-driven decisions. Image: Unsplash — a close up of text on a piece of paper

If you work in the Workers ecosystem or compile to Wasm, watch how the project evolves: that's the kind of environment where computing cost and verification speed matter most. For everyone else, waiting for the repository to mature is a perfectly reasonable option.

Opinion · Open question

The story mentions that code agents can check their own work without exhausting available resources, but it leaves something essential hanging: how to measure that an agent-generated port behaves the same as the original compiler in rare cases, not just common ones.

It matters because, if verification is also done by agents, the risk of subtle errors shifts to the test suite. Before adopting anything, I want to know what real coverage supports the equivalence and who audits it.

Related articles from LaRebelión

Guide: Prepare Your Android UI for AI Agents with Jetpack ComposeHow to use AI to create a Linux driver from a Windows oneIntegrate DuckDB into Amazon Aurora PostgreSQL with This Technical Guide

• AI model recommender for developers
• How to Evaluate Pyrefly in Your Python Project Before Dropping Pytype or Mypy
• Guide to Migrating to Final Structured Concurrency in JDK 28

Original source: thenewstack.io

Article generated with AI.larebelion

Kernel · Head of Technology · Spain

«A compiler written by agents is only as good as the tests that monitor it, and that's what needs to be looked at before speed.»