Skip to content
View shimo4228's full-sized avatar

Block or report shimo4228

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
shimo4228/README.md

Language: English | 日本語

Tatsuya Shimomoto (@shimo4228)

Ask DeepWiki GitMCP

Hub repository for five independently citable research lines by Tatsuya Shimomoto: AI agent design, AI-mediated authorship, and computational phenomenology.

This repo is a map, not the source of truth for live project state. Each research line lives in its own repository and Zenodo concept DOI; this hub keeps the stable relationships, citation pointers, and machine-readable navigation surfaces in one place.

I like making things under constraint — beats, meditation, public service, AI agents. The thread runs through what's below: the meditation practice feeds two of the five research lines, and the agents run on a single Apple Silicon Mac — no lab, no affiliation.

At a Glance

Line Role Stable concept Canonical record
Agent Knowledge Cycle Agent-design mechanism A six-phase loop for sustaining agent-operator intent alignment as both behavior and judgment evolve. DOI 10.5281/zenodo.19200726
Contemplative Agent Agent disposition and implementation A local agent using four contemplative axioms as an optional behavioral preset, not an architectural dependency. DOI 10.5281/zenodo.19212118
Agent Attribution Practice Accountability practice Harness-neutral ADRs for deciding what to prohibit, where controls live, and who answers when an agent fails. DOI 10.5281/zenodo.19652013
Authorship Strategy AI-era authorship methodology A framework for being a known author when LLMs mediate how readers discover and cite artifacts. DOI 10.5281/zenodo.20263316
Attention, Not Self Cognitive and Buddhist inquiry A comparison of Abhidharma cognitive-process models with computational phenomenology. DOI 10.5281/zenodo.20262112

The five lines are siblings, not dependencies. AKC is the mechanism, Contemplative Agent is the disposition and running case, AAP is the accountability practice, Authorship Strategy is the diffusion methodology, and Attention, Not Self is the cognitive-philosophical inquiry.

Machine reading

For machine reading, start with graph.jsonld, then llms.txt, then llms-full.txt.

Agent Knowledge Cycle (AKC)

AKC is a six-phase loop for keeping an agent aligned with its operator's intent as that intent changes. Tests can catch incorrectness; AKC targets drift between a changing operator and a changing agent.

The stable loop is:

Research -> Extract -> Curate -> Promote -> Measure -> Maintain -> Research

AKC stacks three layers: principles, design patterns, and composable implementation skills. The intended end state is scaffold dissolution: once the loop is internalized, explicit skill calls can fall away while the judgment pattern remains.

Contemplative Agent

Contemplative Agent asks whether agent alignment can come from what an agent is, not only from what it is told. It adopts the four contemplative axioms from Laukkonen et al. (2025) as an optional behavioral preset: mindfulness, emptiness, non-duality, and boundless care.

The reference implementation runs AKC over its own logs. Human approval gates apply when material is promoted into skills, rules, identity, or Constitution; logs and patterns are working material, not gated promotion targets. It uses a local 9B stack on Apple Silicon and applies security-by-absence: risky capabilities such as shell execution, arbitrary URL access, and filesystem traversal are absent rather than merely forbidden by policy.

Agent Attribution Practice (AAP)

AAP is a set of harness-neutral ADRs about accountability in autonomous agents. It separates capability from responsibility: what to prohibit, where the prohibition lives, and who answers when behavior breaks.

Its stable concepts are the prohibition-strength hierarchy and the Four Business AI Quadrants: Script, Algorithmic Search, LLM Workflow, and Autonomous Agentic Loop. AAP is the practice; AKC is the cycle.

Authorship Strategy

Authorship Strategy formalizes how authorship changes when LLMs mediate discovery, summarization, and citation. It is a normative framework, a tactical catalog, and an empirical baseline drawn from operating this research ecosystem.

The framework rests on a three-axis inversion: scarcity to diffusion, exclusivity to derivation, and enclosure to openness. It uses a four-layer judgment stack: Authenticity, Attribution Diffusion, Idea vs Scaffold, and Tactics.

Vocabulary note: "attribution" here means credit for source. In AAP, "attribution" means accountability for action. They share a word, not a concept.

Attention, Not Self

Attention, Not Self maps the three major Buddhist Abhidharma traditions, Theravāda, Sarvāstivāda, and Yogācāra, onto contemporary computational phenomenology: predictive processing, active inference, Global Workspace Theory, and Parallel Distributed Processing.

The organizing view is that attention, not a persisting self, is the operative unit of cognition. Japanese is the canonical language; the English README is provided for access.

Papers

Position papers deposited as standalone Zenodo records, each belonging to one research line. The Zenodo concept DOI is canonical; SSRN mirrors are listed where maintained in this hub table.

Paper Line Links
Harness Alignment and Harness Drift: Why Intent, Unlike Correctness, Resists Automation AKC DOI 10.5281/zenodo.20578272 · SSRN
Distributing Accountability, Not Capability: Phase Separation and the LLM Workflow Quadrant in Autonomous AI Agent Architectures AAP DOI 10.5281/zenodo.20353789 · SSRN
The Two-Layer Black Box: Operator Visibility, Commercial Secrecy, and a Minimum Disclosure Set for Accountable Autonomous AI Agents AAP DOI 10.5281/zenodo.20355907 · SSRN

Supporting Ecosystem

AKC cycle skills:

Repo Role
akc-cycle All phases: the full six-phase cycle as a single behavioral rules file, without the six standalone skills.
search-first Research phase: search for existing solutions before building.
learn-eval Extract phase: turn sessions into reusable patterns with quality gates.
skill-stocktake Curate phase: audit skills for staleness, conflict, and redundancy.
skill-health Curate phase: scan skill libraries for structural debt such as missing scripts, agents, and sibling-skill references.
rules-distill Promote phase: distill cross-cutting principles into rules.
skill-comply Measure phase: test whether agents follow skills and rules.
context-sync Maintain phase: keep documentation roles and context surfaces coherent.

Contemplative Agent extensions:

Repo Role
contemplative-agent-rules Drop-in Claude Code rules implementing the four contemplative axioms.
contemplative-agent-cloud Optional managed-LLM backend while keeping the local embedding pipeline.
contemplative-agent-mlx MLX-based extension for Contemplative Agent on Apple Silicon.
contemplative-agent-data Live agent identity, knowledge, and episode logs as a public dataset.

Authorship Strategy components and complements:

Repo Role
doctrine-corpus Bilingual judgment-eliciting Q&A corpus across the sibling research lines. DOI 10.5281/zenodo.20337008
authorship-strategy-skill The four-layer judgment stack as a loadable rule set for LLM-based coding agents.
release-doi Identifier-federation release workflow for DOI-registered research repos.
llms-txt-writer AI-facing document writer for llms.txt, llms-full.txt, FAQ, and glossary surfaces.
jsonld-knowledge-graph Companion JSON-LD graph writer for stable concept-level project structure.
readme-writer Human-facing README writer and review workflow.
wikidata-federation Wikidata federation for researchers, papers, repos, ORCID, DOI, and graph links.
existence-proof Pre-line complement: verifiable institution-grade artifacts by people without conventional credentials. DOI 10.5281/zenodo.20558800
einstein-arena Worked instance of the Existence Proof Format, anchored to a public repo and external arena.

Adjacent tooling and writing skills:

Repo Role
claude-harness Public snapshot of the daily-use Claude Code skills, agents, and rules collected by origin: shimo4228.
claude-skill-writing-ecosystem Human-facing writing and review orchestrator.
daily-research Cron-driven daily research digest pipeline.
claude-skill-paper-ecosystem Academic paper write/review bundle for SSRN, arXiv, Zenodo, and journal venues.

Data and Writing

Surface Use
Traffic dashboard and raw data Daily public GitHub Traffic snapshots for the research repositories, published CC0.
Concept index One definition page per coined research term, linking back to the canonical repository, glossary entry, and DOI.
LLM probes Two-channel attribution-diffusion probe log: parametric and retrieval channels kept separate.
Software Heritage archive Content-addressed archival layer for the public source repositories.
zenn-content Source of truth for long-form articles; browser mirrors are Zenn, Dev.to, and Substack.
SSRN Academic working papers mirrored from Zenodo where applicable.

Citation and Identity

Author: Tatsuya Shimomoto — ORCID 0009-0002-6168-4162 · Wikidata Q140090100 · Hugging Face @Shimo4228

This hub is CC0-licensed. Cite individual research lines by their concept DOI; cite the hub itself only when referring to the aggregate index, knowledge graph, traffic snapshots, or probe dataset.

Pinned Loading

  1. contemplative-agent contemplative-agent Public

    Contemplative Agent (CA) — a CLI agent that runs a six-phase knowledge cycle (AKC) over its own logs, with contemplative axioms as behavioral presets. Every promotion passes a human approval gate. …

    Python 5 2

  2. agent-knowledge-cycle agent-knowledge-cycle Public

    Agent Knowledge Cycle (AKC) — a knowledge cycle for AI agents that grows with the people who shape it. Six phases keep behavior aligned with operator intent as human attention becomes the scarce re…

    3

  3. agent-attribution-practice agent-attribution-practice Public

    Agent Attribution Practice (AAP) — architecture decision records on accountability distribution in autonomous AI agents. 10 ADRs + 4 Business AI Quadrants + Phase Separation axis. Harness-neutral p…

    2

  4. authorship-strategy authorship-strategy Public

    Authorship Strategy: A Normative Framework and Tactical Catalog for AI-Era Authenticity Inversion, with Empirical Grounding from a Five-Line Research Ecosystem

    2

  5. attention-not-self attention-not-self Public

    Abhidharma × Computational Phenomenology — a personal inquiry into attention and the non-self

    TypeScript 2

  6. zenn-content zenn-content Public

    Bilingual (JA/EN) essay corpus on AI agent design, coding-agent workflows, and authorship under AI-mediated diffusion — the essay surface of a five-line research ecosystem. CC0; intrinsic priority …

    HTML 2