A large claim, bounded by public evidence.
Millrace autonomously built its Rust port and now maintains that port through a public evidence loop. This page states what that proves, what it does not prove, and where to inspect the artifacts.
The first campaign is inspectable.
The initial Rust port campaign is documented through public summaries and artifacts. The important point is not that the run was magic; it is that the bounded system produced a second implementation with visible planning, task, run, and result evidence.
Campaign metrics: 8 seeded parity slices/ideas, 11 completed specs, 57 completed tasks, 99 runs, 261 stage calls/results, 28h 9m 49.5s campaign span, and 730M input plus output tokens.
The evidence rail stays bounded.
The campaign evidence lives at github.com/tim-osterhus/millrace-rs-port-docs, including the claim framing in CLAIMS.md.
The maintenance claim is bounded to the public loop that keeps the Rust port aligned after Python releases, with the operator-facing surface visible at live.millrace.ai.
A second proof rung is documented at github.com/tim-osterhus/millracer-rs-port-docs, covering the downstream Millracer port.
The evidence does not prove universal software autonomy, zero human involvement, or a complete forensic replay. The public ledger and claim document are the right place to inspect where the claim starts and stops.