Cloudflare has announced that it is taking over the development of the tsrs project, a TypeScript compiler written in Rust. The project aims to address performance challenges faced by AI Agents when checking large codebases, as initial tests showed the new compiler is eight times faster and consumes 70% less memory compared to TypeScript 7.0.
Not a Fork, but an Integration
Cloudflare engineers, led by Christoph Nakazawa, emphasized that this project is not a "fork" of the original TypeScript language. The team explained that they are working in coordination with the TypeScript team at Microsoft, noting that improvements discovered in the tsrs project have already contributed to accelerating the TypeScript 7.1 release by approximately 2.5 times.
Why Rust?
Although Microsoft recently migrated the TypeScript compiler to Go, Cloudflare sees a competitive advantage in Rust, particularly regarding compatibility with WebAssembly (Wasm). While Go requires a Garbage Collector when compiling to Wasm, Rust provides direct memory control, reducing operational overhead when running development environments inside browsers or in isolated environments like Cloudflare Workers.
The Origin Story: Development "Without Human Code"
The tsrs project began at Owner.com, where Max Shenk, Head of AI Developer Experience, faced difficulties dealing with a codebase containing 37,000 files. To solve this problem, Shenk tasked AI agents (using Claude Opus 5.5) with rewriting the compiler entirely in Rust, without him writing a single line of code. This approach reduced code checking time in Continuous Integration (CI) environments to just about 7 seconds.
What Does This Mean for Developers?
Developers face an ongoing challenge in ensuring the new compiler's compatibility with original TypeScript standards. To avoid behavioral drift, the team relies on:
- Using TypeScript's own conformance tests.
- Injecting intentional errors into real projects to verify checking accuracy.
- Using the Miri tool to detect memory safety issues in Rust.
Cloudflare plans to integrate tsrs into its Oxc toolchain, while continuing to monitor upcoming updates from Microsoft to ensure full compatibility, reflecting a growing trend toward using AI to build high-performance engineering tools.