1\section*{AI-use and disclosure statement}2\addcontentsline{toc}{section}{AI-use and disclosure statement}34This study was conceived, specified and reviewed by the named author and executed by AI systems; the division of labour is stated here because the paper's subject is that division. The research mandate (a 6,900-word instruction file, preserved verbatim with its hash in the repository) was prepared by Simon-Pierre Boucher outside any AI session; its preparation time was not measured and is reported as unknown. The mandate was pasted into one Claude Code session (client 2.1.314, model \texttt{claude-fable-5-1}, Anthropic) with a four-word instruction to carry out the project. That session, called the builder in the text, designed and implemented the harness, wrote the protocols, the smoke task, the analysis pipeline and all sections of this manuscript except the related-work and corpus sections, ran every experiment, and delegated to fresh sub-sessions of the same client (model \texttt{claude-fable-5-1}) the corpus coding, the literature review, the authoring of the six task families and the three transfer projects with their hidden oracles, the drafting of the related-work and corpus sections, and three adversarial reviews (systems and isolation; statistics and evaluators; manuscript claims). The experimental subjects were \nAllRuns{} fresh Claude Code sessions (client 2.1.314, model \texttt{claude-opus-5-5}, Anthropic) in bare mode on the author's cluster nodes; the arm-D reviewers were sessions of the same model. All outcome evaluation was performed by deterministic Python programs written by the task-authoring sub-sessions and validated on seeded artefacts; no language model graded any artefact for the reported endpoints. Model identities are provenance, not authorship: the named author is responsible for the mandate, for the decision to run the study, and for the final review of the manuscript, which is logged in \texttt{reports/interventions.csv} with its date and basis. No human intervened in any executor session. The literature review used Crossref, arXiv, Semantic Scholar and OpenReview records retrieved by a sub-session; every reference was verified against those records and every DOI was checked to resolve. No funding, grant, institutional approval or human participant is associated with this work. Venue-specific disclosure requirements will be checked before any submission; no submission has been made.5