\clearpage
\begin{appendices}
\section{Task suite and task cards}
\label{app:tasks}
\input{tables/tab_tasks}
\input{appendices/task_cards}
\clearpage
\section{Treatment texts (verbatim)}
\label{app:prompts}
The common preamble is prepended to every arm text and injected with \texttt{--append-system-prompt-file}; the user prompt is identical for all arms. SHA-256 hashes of every file are in \texttt{protocol/freeze\_manifest\_conf.json}.
\subsection*{Common preamble}
\VerbatimInput[fontsize=\scriptsize,breaklines=true]{../protocol/arms/common_preamble.md}
\subsection*{Arm A}
\VerbatimInput[fontsize=\scriptsize,breaklines=true]{../protocol/arms/arm_A.md}
\subsection*{Arm B}
\VerbatimInput[fontsize=\scriptsize,breaklines=true]{../protocol/arms/arm_B.md}
\subsection*{Arm C (text appended to B)}
\VerbatimInput[fontsize=\scriptsize,breaklines=true,firstline=38]{../protocol/arms/arm_C.md}
\subsection*{Arm D (text appended to C)}
\VerbatimInput[fontsize=\scriptsize,breaklines=true,firstline=43]{../protocol/arms/arm_D.md}
\subsection*{Reviewer (arm D)}
\VerbatimInput[fontsize=\scriptsize,breaklines=true]{../protocol/arms/reviewer.md}
\subsection*{User prompt, gate feedback, review feedback and resumption prompts}
\VerbatimInput[fontsize=\scriptsize,breaklines=true]{../protocol/arms/task_prompt.md}
\VerbatimInput[fontsize=\scriptsize,breaklines=true]{../protocol/arms/gate_feedback_prompt.md}
\VerbatimInput[fontsize=\scriptsize,breaklines=true]{../protocol/arms/review_feedback_prompt.md}
\VerbatimInput[fontsize=\scriptsize,breaklines=true]{../protocol/arms/resume_after_interruption_prompt.md}
\subsection*{Length-matched control (arm L)}
\VerbatimInput[fontsize=\scriptsize,breaklines=true]{../protocol/arms/arm_L.md}
\clearpage
\section{Evaluator validation}
\label{app:evaluators}
\input{appendices/evaluator_validation}
\clearpage
\section{Statistical details and additional results}
\label{app:stats}
\input{appendices/stats}
\clearpage
\section{Environment and model records}
\label{app:env}
\input{appendices/environment}
\clearpage
\section{Intervention and amendment ledgers}
\label{app:ledgers}
\input{appendices/ledgers}
\clearpage
\section{Reproduction instructions}
\label{app:repro}
\input{appendices/repro}
\clearpage
\input{appendices/gallery}
\end{appendices}