Modern LLM systems reset cognitive state between sessions, breaking continuity of human-directed workflows. SUCC–ECC is a system-level architecture for persistent, human-directed cognitive continuity using externally stored, reproducible memory artifacts.
This is not model-level memory tuning (no weight updates or parameter modification). It is a structured operational continuity framework using externally stored, reproducible memory artifacts.
- Defines SUCC: governance and roles for human-AI co-evolution
- Defines ECC: operationalized external memory layer
- Demonstrates persistent continuity without internal model modification
Long-term human-AI collaboration across iterative research, writing, and system design cycles.
Author: Seungyun Song
Year: 2025
Status: Independent research
Paper DOI: https://doi.org/10.5281/zenodo.17778738
Software / Repository DOI: https://doi.org/10.5281/zenodo.18586738
ORCID: https://orcid.org/0009-0001-2177-0042
LinkedIn: https://kr.linkedin.com/in/seungyun-song-735395393
Summary post(Medium): https://medium.com/@ssp2025.research/human-directed-human-ai-co-evolution-through-externalized-cognitive-continuity-succ-445f77e6b37d
This repository serves as a reference companion to the paper.
Status: Reference materials are available; implementation artifacts will be added in future updates.
“Operational framework with a reference implementation”
SUCC-ECC is a system-level architecture for persistent human–AI co-evolution through externalized cognitive continuity (ECC: Externalized Cognitive Continuity), where SUCC defines the human-centered framework and ECC provides operational continuity via externally stored, human-controlled cognitive artifacts.
Note: This repository does not introduce a new conceptual framework separate from the paper. All terminology and system boundaries are authoritatively defined by the associated Zenodo preprint (2025).
⚠️ Note on Rule Design and StabilityThis framework assumes layered rule design, not flat rule lists.
In practice, most instability and unexpected behavior arise not from rule content,
but from rule placement and priority conflicts.Rules should therefore be:
- structured in explicit layers,
- assigned clear precedence,
- and validated iteratively in interaction with the model, not written unilaterally.
SUCC-ECC is a system-level architecture that integrates
SUCC (Syntropic Unified Cognitive Continuum) and
ECC (Externalized Cognitive Continuity) as complementary components.
It is not a subcategory of ECC.
Rather, ECC functions as the operational memory layer within the SUCC-directed system.
SUCC defines governance, roles, and workflows.
ECC implements reproducible, externalized memory substrates.
Together, they form SUCC-ECC, enabling persistent human–AI co-evolution without relying on internal model state or fine-tuning.
Terminology note
SUCC-ECC is the author-defined system architecture introduced in this paper. In the paper, SUCC is presented as the framework and ECC as the external memory architecture; this repository uses "SUCC-ECC" as a convenient compound term for the integrated system described in the paper.
The phrase “Externalized Cognitive Continuity” is introduced in the title of the associated Zenodo preprint (2025) to denote a specific continuity mechanism realized through externally stored, human-controlled cognitive artifacts.
It is not a general term, nor a sub-framework of any external ECC taxonomy.
All references to “SUCC”, “ECC”, and “SUCC-ECC” in this repository follow the definitions established in the associated Zenodo preprint (2025).
Human-directed human–AI co-evolution enabled by externalized cognitive continuity (logs, rules, and structured re-injection), providing a practical time-axis for iterative, human-centered collaboration.
To achieve human goals,
through a human-directed and structured approach,
the conversational identity of AI is reconstructed,
enabling AI assistance
toward human cognitive insight and tangible human flourishing.
This work introduces SUCC (Syntropic Unified Cognitive Continuum) as a human-directed Human–AI co-evolution framework designed to enable long-term cognitive continuity across otherwise stateless interaction sessions.
Within the SUCC framework, ECC (Externalized Cognitive Continuity) serves as the core architectural mechanism that operationalizes continuity through user-controlled external artifacts. These include dialogue logs, explicit rule sets, temporal checkpoints, and structured re-injection procedures that reconstruct stable cognitive context at session initialization.
In short:
- SUCC specifies the framework-level principles, roles, and iterative processes governing sustained Human–AI co-evolution.
- ECC provides the architectural implementation that realizes cognitive continuity through externalized, reproducible memory structures.
- SUCC–ECC System Architecture Overview: ARCHITECTURE.md
- DOI landing page: https://doi.org/10.5281/zenodo.17778738
- Direct PDF: https://zenodo.org/records/17778738/files/Song2025_SUCC_ver3.pdf?download=1
Please use the following citation when referencing the paper.
Song, S. (2025). Human-Directed Human-AI Co-Evolution through Externalized Cognitive Continuity. Zenodo. https://doi.org/10.5281/zenodo.17778738
This GitHub repository is archived and versioned via Zenodo.
-
Concept DOI (all versions, resolves to the latest): https://doi.org/10.5281/zenodo.18586738
-
Versioned releases:
This repository contains two distinct components with different licenses:
The paper is hosted on Zenodo and is not part of this repository’s contents.
- Licensed under CC BY-NC-ND 4.0
- https://creativecommons.org/licenses/by-nc-nd/4.0/
All materials in this GitHub repository (framework documentation, architectural diagrams, and related materials) are licensed under:
- CC BY-NC 4.0
- https://creativecommons.org/licenses/by-nc/4.0/
If referencing this work academically, please cite the Zenodo DOI provided above.
@misc{song_2025_17778738,
author = {Song, Seungyun},
title = {Human-Directed Human-AI Co-Evolution through Externalized Cognitive Continuity},
month = nov,
year = 2025,
publisher = {Zenodo},
doi = {10.5281/zenodo.17778738},
url = {https://doi.org/10.5281/zenodo.17778738},
}