The Mekiki Framework is a research program by Kengo Tomita on what can and cannot be delegated in AI-assisted knowledge work. Papers T1–T4 decompose the boundary of legitimate AI delegation; T5 asks the concluding converse question — why an agent may rationally decline delegation even when delegation is technically possible.

• Website: https://mekiki-framework.github.io/
• Theory Map: https://mekiki-framework.github.io/THEORY_MAP.md
• Consolidated Markdown: https://mekiki-framework.github.io/mekiki-framework-t1-t5.md
• AI entry points: https://mekiki-framework.github.io/llms.txt · https://mekiki-framework.github.io/FOR_AI_READERS.md

• T1: https://doi.org/10.31235/osf.io/cwkav_v1
• T2: https://doi.org/10.31235/osf.io/e9qw5_v2
• T3: https://doi.org/10.35542/osf.io/hvbfe_v2
• T4: https://doi.org/10.31235/osf.io/495wg_v1
• T5: https://doi.org/10.31235/osf.io/593ah_v1

Status (August 2026): T1 is a preprint whose journal version is under peer review; T2–T5 are preprints.

The website and repository are derivative access layers. Cite the relevant paper DOI as the scholarly source.

Author: Kengo Tomita, Institute of Technology, Shimizu Corporation
ORCID: https://orcid.org/0000-0002-7310-8093