On October 20, 2026, Cloudflare and VoidZero announced that, with Max Schwenk of Owner.com, they are working on tsrs, "a Rust port of the TypeScript compiler", incubated in the Oxc toolchain "in close collaboration with the TypeScript team." The post reached 349 points on Hacker News.
Why another port
The context is what makes this a trend. Cloudflare writes that "frontier AI models can now do large-scale ports almost independently in a matter of days." It continues: "We have now seen slopforks, faithful ports, and completely re-architected versions of TypeScript in Rust or C++." Its pitch for tsrs is restraint: "To be clear, we are not forking TypeScript," and a port "has to follow those decisions closely, or it will eventually fall apart and become incompatible." Early benchmarks, run in a configuration agreed with the TypeScript team, put tsrs at "up to 8x faster" with "~70% less memory" than TypeScript 7.0 at equal worker counts. Some optimisations have already been upstreamed into the 7.1 beta, which Cloudflare says is up to 2.5x faster than 7.0. The main reason to use Rust over Go is WebAssembly: there is no garbage collector to ship, so a complete toolchain can run in a sandbox, a browser or an embedded device.
Agents as the sync mechanism
The maintenance plan copies Cloudflare's Vinext, its Next.js reimplementation: "Each morning, an agent reviews the changes, fetches diffs, and opens tracking issues". Every night the upstream test suite runs, and agents "identify the change across both codebases, build a reproduction, port any relevant tests, and propose a fix." The tsrs README sets 100% conformance with TypeScript 7.1 as the bar for a stable release. It lists known differences, such as diagnostic ordering, and says "Pull requests remain closed because we don't have the bandwidth to review them." Cloudflare does not say how much of tsrs itself was written by models.
The same pattern elsewhere this week
DeepSeek's DeepInfer, covered in today's lead entry, says it "was originally built primarily on vLLM" before becoming the production engine for V4.1 Flash. The upstream continued as a starting point, and the derived engine became the product. The human side of the shift showed up on October 19 and 20. Simon Willison highlighted Jeremy Scheff's "My obituary for programming", which describes being paid "to work on tricky little logic puzzles"; Willison wrote that for people who enjoyed that part, "the loss is very real." In "What Shall I Teach My Son?", Steve Krouse, an expectant parent, writes: "My economically valuable skillset – programming – suddenly feels less valuable." The essay drew 342 comments on Hacker News on October 20.
What remains uncertain. tsrs is experimental, its speed figures are self-reported, and it is unclear how an agent-driven sync loop will hold up when TypeScript makes a large design change that takes years to settle.
Analysis: When a rewrite costs days rather than years, a port is no longer scarce, and what becomes valuable is evidence that it behaves exactly like the original: shared tests, agreed benchmark configurations, documented differences and an upstream that keeps control of the design. Teams considering an agent-driven port of their own systems should budget for that layer first. That means a conformance suite, a daily diff against upstream and a named owner for divergence. Without it, the port becomes a slopfork once the original moves on.
Cloudflare and VoidZero's tsrs (October 20), a Rust port of the TypeScript compiler built with the TypeScript team's support, claims up to 8x speed and ~70% less memory than TypeScript 7.0 and plans to stay in sync through daily agent-driven diffing and nightly upstream tests, after Cloudflare noted frontier models can now port large codebases in days; with DeepInfer grown from vLLM, the scarce part of software is shifting from writing a port to proving it stays faithful.