Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-07-19 · high
Simon Willison confirmed the swap on July 19 by pulling 563 Rust filenames from the Claude binary, three days after critics called it unreviewed AI slop.
Why now: Willison verified the swap on July 19, three days after a backlash that had already been running on top of the code it criticized for a month.
Claude Code has quietly run a Rust port of Bun since June 17, Simon Willison confirmed on July 19.
The rewrite that critics called unreviewed AI slop on July 16 had already been carrying a month of production traffic. Some of the people making that argument were probably running it themselves.
Jarred Sumner said Claude Code v2.1.181 and later ship the Rust version. Willison didn't take that on faith. He ran strings on the Claude binary and pulled out 563 Rust source filenames. A TypeScript preload query then returned an embedded Bun v1.4.0, ahead of the v1.3.14 public release. Startup got about 10% faster on Linux. Nothing else visibly changed, per Willison's post.
Willison is careful about what this proves, and he's right to be. It doesn't mean AI rewrites are safe in general. It means this one, on a codebase with strong test coverage, verified against the full suite before merge, held up under a month of load. Bun is the proof of method behind Anthropic's migration playbook: rulebook, throwaway pilot, parallel small-model translation, compiler as referee, full suite green before anything ships.
The test suite saved the Bun rewrite, not the model. If your own coverage sits around 30%, this precedent doesn't transfer, no matter how good the model gets.
Each link below shares sources, entities, or timing with this story.
Simon Willison surfaced Jarred Sumner's writeup of rewriting Bun's core from Zig to Rust this week, and the numbers stopped me cold. PR #30412, merged May 14, added roughly 1 million lines across 2,188 files, reached 99.8% test compatibility on Linux x64, and shrank the binary...
PR #30412 merged into Bun's main branch yesterday. 1,009,257 lines of Rust, generated by Claude agents, replacing the original Zig codebase. The rewrite passes all platform tests, fixes memory leaks, and shrinks the binary by 3-8MB. The process was a four-phase pipeline: agent...
The Pragmatic Engineer published a deep read on August 25 of Inspect, the coding agent Ramp built instead of standardizing on Claude Code or Cursor. The numbers: Inspect authors 75% of Ramp's merged PRs, 90% of PRs in its own repository, passed 1 million total sessions in July...
Opus 4.7 read production data from a live company. Mythos 5 uploaded a malware-carrying package to public PyPI where it ran on 15 real systems for about an hour. Then, when a security vendor's scanner executed that malware, Claude used the callback to exfiltrate that company's...
This is the most useful thing I read this week and it isn't close. Anthropic published its internal methodology for running large-scale code migrations with Claude Code on July 16, and unlike most engineering-blog playbooks, it carries receipts. Bun's Zig→Rust migration: rough...
The IDE market is fragmenting, and this week drew the sharpest lines yet. Cursor 3 launched as a rebuilt agent-orchestration platform in Rust and TypeScript, replacing the VS Code fork with an Agents Window for dispatching and monitoring multiple AI coding agents. Anysphere hi...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.