\section{Reproducibility and execution disclosure}
\label{sec:repro}

\paragraph{How the study was executed.} All code, experiments, analyses and this manuscript were produced by Claude Code (Anthropic;
model \texttt{claude-fable-5-1}) acting as research operator under a written brief from the author, through actual terminal
commands on the author's machines: environment inspection, dependency pinning, implementation, launching inference, reading exit
codes and logs, diagnosing failures, re-running affected trials, analysis and compilation. The author provided the brief, the
hardware, API credentials already present in sibling projects, and no mid-run intervention; dated engineering decisions are in
\texttt{DECISIONS.md}. Models under test are the experimental subjects; the operator's own inference state was never an
experimental interface. This disclosure describes the process; it is not a claim that autonomous research is superior to any
alternative.

\paragraph{Defects found and handled.} (i) The \texttt{mlx-lm} cache API changed between the system-installed version (0.31.3) and the
pinned environment (0.32.2); adapters were rewritten for the pinned API before any experiment. (ii) The receiver prefix boundary was
initially computed from the prefix text alone and did not match the tokenization of the full prompt; it is now the common token
prefix with a canonical question, and a mismatch raises. (iii) Heterogeneous retained lengths (B6) broke the shared attention mask;
the fresh query is processed token by token for such caches. (iv) The first executable-task specification (seven rules) was unsolvable
for 3B receivers even with the full ticket; it was simplified on development data before any test item was run (D-010). (v) Running
three GPU jobs concurrently exhausted unified memory and slowed the pilot by an order of magnitude; GPU jobs were serialized.

\paragraph{Artifacts.} The repository contains the harness, pinned lock file, data manifests and generators, the frozen protocol and its
hash, every per-attempt record (\texttt{runs/raw}), sanitized logs, the mapper checkpoint and manifest, generated tables and figures
with a provenance index linking each to its source records and script, a claim ledger, and the minimal reproduction profile.
Checkpoints are not redistributed; their Hugging Face identifiers, revisions and weight hashes are recorded. Hosted-API trials require
the provider credentials and reproduce only up to provider nondeterminism.

\paragraph{Reproduction commands.}
\begin{footnotesize}\begin{verbatim}
make reproduce-analysis      # tables + figures from runs/raw, provenance audit
python -m wp30 reproduce --profile reproduction/minimal.yaml
                             # re-run a fixed subset, compare greedy outputs
make paper                   # compile this document
\end{verbatim}\end{footnotesize}
Greedy decoding on MLX is deterministic for an identical prefill chunking but not bitwise across chunkings (Section~\ref{sec:correctness});
the reproduction report lists identical output hashes and identical scores separately.