curl's Stenberg: agents rewrote 185,000 lines in Rust, curl fixed 200+ bugs

Simon Willison's blog on 18 November 2026 quotes a short passage from Daniel Stenberg's article "Actual curl rewrites in Rust". The post is only the quote and its attribution.

Stenberg sets two streams of work side by side. During the period in which agents converted 185,000 lines of C code into 340,000 lines of Rust code, a figure he says he counted himself, "we merged over 300 commits in the actual curl project". He says the period was a little more than three weeks, and that curl was actually improved in that time.

The improvements were concrete: a little over 200 bugfixes. Stenberg's point is that these are 200 bugfixes that none of those Rust rewrites have. The curl project also fixed a few vulnerabilities in the same period.

He then looks ahead. In another month, he expects, curl will have done another 200 bugfixes and maybe fixed a few more vulnerabilities. The passage ends on a question: "Will these clones keep up?"

The excerpt does not name the agents, the companies or the people behind the Rust rewrites, and it does not say how many rewrites there are or where they are published.

Key facts

  • Stenberg says agents converted 185,000 lines of C into 340,000 lines of Rust, and adds "(I counted)".
  • In the same period of a little more than three weeks, the actual curl project merged over 300 commits.
  • Those commits included a little over 200 bugfixes, which Stenberg says none of the Rust rewrites have, plus a few vulnerability fixes.
  • He expects another 200 bugfixes and maybe a few more vulnerability fixes in the next month, and asks: "Will these clones keep up?"
  • The source is a quote on Simon Willison's blog, dated 18 November 2026, taken from Stenberg's article "Actual curl rewrites in Rust".

Why it matters

Rewrites of old C code in Rust by agents are easy to judge by size: lines in, lines out. Stenberg offers a different yardstick, the steady stream of fixes that a living project collects. In his telling, a conversion that took 185,000 lines of C to 340,000 lines of Rust happened while curl itself moved on with over 300 commits and a little over 200 bugfixes. The comparison puts the cost of staying current at the centre of the argument.

Who it affects

The curl project, whose contributors Stenberg speaks for when he says "we merged". Also anyone who is weighing an agent-made Rust rewrite of a C codebase, since his question is whether a clone can keep pace with the original's fixes. The excerpt does not name the agents or the people behind the rewrites.

How to use it

There is nothing to install or run here; the post is a quoted opinion with counts. The practical takeaway is the comparison itself: when you look at a rewrite, ask what bugfixes and vulnerability fixes the original project has landed since the conversion, and whether the rewrite has them.

How solid is it

The line counts (185,000 lines of C, 340,000 lines of Rust) are Stenberg's own, marked "(I counted)". The commit and bugfix numbers describe his own project, and he gives them as "over 300" and "a little over 200". The claim that none of the rewrites have those 200 fixes is his assertion. The excerpt does not give the method of counting, and it does not give the start or end dates of the three weeks. The forecast of another 200 bugfixes in a month is an expectation, worded with "maybe" for the vulnerabilities.

Risks and caveats

This is a short excerpt, not the full article. It does not say whether the Rust rewrites work, are complete or tested, and it gives no quality or performance comparison. It does not say how many vulnerabilities were fixed ("a few") or how severe they were. The framing is Stenberg's, and Simon Willison adds no commentary of his own. Raw counts of commits and bugfixes also measure activity, not the value of each change.

“We actually improved curl during these a little more than three weeks. A little over 200 bugfixes actually. 200 bugfixes that none of those rust rewrites have.”

— Daniel Stenberg, Actual curl rewrites in Rust