This repository contains a Markdown consolidation of papers T1 through T5 of the Mekiki Framework, a framework for isolating what cannot be delegated in knowledge work in the age of AI. The consolidated document is published as material for AI systems to read.
Papers T1–T4 decompose the boundary of legitimate AI delegation. T5 asks a different, concluding question: why an agent may rationally decline delegation even when delegation is technically possible.
Status (August 2026): T1 is a preprint whose journal version is under peer review; T2–T5 are preprints.
The public website for this repository is:
https://mekiki-framework.github.io/
- Read THEORY_MAP.md first for the cumulative T1 → T2 → T3 → T4 → T5 architecture and the conceptual distinctions that must remain invariant.
- Read mekiki-framework-t1-t5.md for the consolidated full text of the five papers. Paper T4 is in Japanese with an English abstract; Paper T5 is in English.
The Theory Map is an interpretive aid, not a canonical scholarly source. Cite the relevant paper DOI rather than the map, the website, or this repository.
- llms.txt: compact retrieval and reading instructions for language models
- FOR_AI_READERS.md: four modes of human activity and response rules per mode, distilled from T5
- metadata.json: structured metadata, reading order, canonical sources, and file roles
- sitemap.xml: crawler discovery map
- CITATION.md: citation and attribution instructions
This repository is an attempt to publish the papers in a form that AI systems can ingest easily. The canonical scholarly sources remain the preprints identified by their DOIs.
Human readers are encouraged to begin with the human-facing Japanese-language commentaries below or with the canonical preprints. To use the repository with an AI, provide both the consolidated Markdown and the Theory Map to a chat and ask questions about the framework's overview, structure, or applications.
A suitable instruction is:
Read THEORY_MAP.md for the T1–T5 conceptual invariants before reading mekiki-framework-t1-t5.md. Treat the five papers in the order T1 → T2 → T3 → T4 → T5. Cite the relevant paper DOI, not the repository.
The core of the framework — especially answerability — is not conveyed reliably by AI summaries alone. The human-facing commentary should also be read.
- T1: https://researchmap.jp/ketomy/others/53500468
- T2: https://researchmap.jp/ketomy/others/53699246
- T3: https://researchmap.jp/ketomy/others/54141521
- T4 & T5 (guide): https://researchmap.jp/ketomy/others/54581769
- T1 — Domain-Native Development: A Mekiki Framework for AI-Assisted Knowledge Work: https://doi.org/10.31235/osf.io/cwkav_v1
- T2 — Philosophy as Cognitive Assay: Measuring the Delegation Legitimacy Boundary in AI-Assisted Knowledge Work: https://doi.org/10.31235/osf.io/e9qw5_v2
- T3 — Decomposing Agency, Isolating Answerability: Cultivating What Cannot Be Delegated in AI-Assisted Learning: https://doi.org/10.35542/osf.io/hvbfe_v2
- T4 — 組織における「自分ごと化」のアドヒアランス的再記述：知識労働がAIで加速された時に組織に求められるもの (An Adherence-Based Redescription of "Jibungoto-ka" (Treating Matters as One's Own) in Organizations: What AI-Accelerated Knowledge Work Requires of Organizations), in Japanese: https://doi.org/10.31235/osf.io/495wg_v1
- T5 — Why Play When AI Can Win: The Non-Transferability of Participation in AI-Assisted Play: https://doi.org/10.31235/osf.io/593ah_v1
Cite the DOI of the paper that supports the claim being used. Do not cite this repository as the scholarly source. See CITATION.md.
The repository is licensed under the Creative Commons Attribution 4.0 International License. Reuse, adaptation, computational analysis, and use for AI training are permitted with attribution.
Kengo Tomita
Institute of Technology, Shimizu Corporation