@id
string
@type
unknown
name
string
description
string
isBasedOn
string
mainEntity
string
contrastsWith
unknown
implements
unknown
about
unknown
alternateName
unknown
siblingOf
unknown
derivesFrom
unknown
author
unknown
groundedIn
unknown
identifier
string
url
string
isPartOf
string
workExample
list
subjectOf
string
definesConcept
list
ordinal
int64
level
int64
extends
string
hasFailureMode
string
appliesTo
string
sameAs
list
datePublished
string
encodingFormat
string
programmingLanguage
string
codeRepository
string
https://shimo4228.github.io/shimo4228/vocab#prior-art/belief-deviation
[ "ExternalReference", "ScholarlyArticle" ]
Meta-Cognitive Memory Policy Optimization for Long-Horizon LLM Agents
One of three further 2026 drift coinages audited in the position paper's pre-deposit sweep (§6): belief deviation over long horizons. Its reference list contains no classical software-evolution literature — part of the disconnection harness drift bridges by reference.
null
null
null
null
null
null
null
null
"Liu, Z. et al."
"https://doi.org/10.5281/zenodo.20578272"
arXiv:2605.30159
https://arxiv.org/abs/2605.30159
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/geo
[ "ExternalReference", "ScholarlyArticle" ]
GEO: Generative Engine Optimization
Measurement framework behind the geo-writer snapshot in ADR-0010 — the first Measure-phase self-application to AKC's own documentation. The README is scored before and after the cognitive-economy change on GEO-derived checks (entity density, question-heading prominence, chunk self-containment, definition density), with...
null
null
null
null
null
null
null
null
"Aggarwal, P. et al."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0010-human-cognitive-resource-as-central-constraint.md"
arXiv:2311.09735
https://arxiv.org/abs/2311.09735
null
null
null
null
null
null
null
null
null
null
2023
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/intrinsic-metacognitive-learning
[ "ExternalReference", "ScholarlyArticle" ]
Truly Self-Improving Agents Require Intrinsic Metacognitive Learning
Taxonomy named in ADR-0005's defense of the human approval gate. In its intrinsic/extrinsic distinction, AKC's gate is extrinsic metacognition — a human-designed loop with a human evaluator at the decision point — and stays so by design, not by immaturity: behavior-modifying writes remain gated because they are where t...
null
null
null
null
"https://shimo4228.github.io/shimo4228/vocab#akc/concept/human-approval-gate"
null
null
null
"Liu & van der Schaar"
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0005-human-approval-gate.md"
arXiv:2506.05109
https://arxiv.org/abs/2506.05109
null
null
null
null
null
null
null
null
null
null
2025
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/externalization-review
[ "ExternalReference", "ScholarlyArticle" ]
Externalization in LLM Agents: A Unified Review of Memory, Skills, Protocols and Harness Engineering
Field map named in ADR-0013's Related-Work positioning. Frames the field as three coupled forms of externalization — memory, skills, protocols — coordinated by harness engineering as the unification layer; AKC accepts that it sits squarely inside this frame, overlapping the memory and skills quadrants, and locates its ...
null
null
null
null
"https://shimo4228.github.io/shimo4228/vocab#concept/three-layer-structure"
null
null
null
"Zhou et al."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2604.08224
https://arxiv.org/abs/2604.08224
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/skillops
[ "ExternalReference", "ScholarlyArticle" ]
SkillOps: Managing LLM Agent Skill Libraries as Self-Maintaining Software Ecosystems
Prior art named in ADR-0013's 2026-06-25 addendum. Frames 'skill technical debt' and a four-dimension library-health diagnosis (utility, compatibility, risk, validation), and proposes a self-maintaining skill ecosystem that prunes, validates, and de-duplicates skills autonomously — the Curate and Maintain operations ru...
null
null
null
null
null
null
null
null
"Pu, Song & Zhao"
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2605.13716
https://arxiv.org/abs/2605.13716
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/skill-lifecycle-sok
[ "ExternalReference", "ScholarlyArticle" ]
SoK: Agentic Skills — Beyond Tool Use in LLM Agents
Field map named in ADR-0013's 2026-06-25 addendum. Maps a seven-stage skill lifecycle (discovery, practice, distillation, storage, retrieval, execution, evaluation/update) for the skill layer the way the Externalization survey maps the broader field. AKC's six-phase cycle overlaps this lifecycle; the delta is not the s...
null
null
null
null
null
null
null
null
"Jiang et al."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2602.20867
https://arxiv.org/abs/2602.20867
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/skillsbench
[ "ExternalReference", "ScholarlyArticle" ]
How Well Do Agentic Skills Work in the Wild: Benchmarking LLM Skill Usage in Realistic Settings
Empirical corroboration named in ADR-0013's 2026-06-25 addendum. Finds that skill benefits are fragile — degrading toward the no-skill baseline as the library grows large and uncurated and the agent must retrieve from many real-world skills rather than be handed a hand-crafted one. Independent support for ADR-0010's cl...
null
null
null
null
null
null
null
null
"Liu et al."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2604.04323
https://arxiv.org/abs/2604.04323
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/sdb-runtime-patterns
[ "ExternalReference", "ScholarlyArticle" ]
A Methodology for Selecting and Composing Runtime Architecture Patterns for Production LLM Agents
Corroboration named in ADR-0013's 2026-07-08 addendum. Names the stochastic-deterministic boundary (SDB) — a four-part contract of proposer, verifier, commit step, and reject signal — as the load-bearing primitive of production agent runtimes, and organizes runtime design into coordination, state, and control concerns ...
null
null
null
null
"https://shimo4228.github.io/shimo4228/vocab#akc/concept/code-llm-layering"
null
null
null
"Srinivasan, V."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2605.20173
https://arxiv.org/abs/2605.20173
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/context-engineering-2
[ "ExternalReference", "ScholarlyArticle" ]
Context Engineering 2.0: The Context of Context Engineering
Corroboration named in ADR-0013's 2026-07-08 addendum. Treats the selection and shaping of what reaches the model as a discipline in its own right, with a history predating LLMs. Independent convergence with signal-first (information that changes no action is not taken in) and cognitive economy (attention managed as th...
null
null
null
null
[ "https://shimo4228.github.io/shimo4228/vocab#akc/concept/signal-first", "https://shimo4228.github.io/shimo4228/vocab#akc/concept/cognitive-economy" ]
null
null
null
"Hua, Q. et al."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2510.26493
https://arxiv.org/abs/2510.26493
null
null
null
null
null
null
null
null
null
null
2025
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/cognitive-divergence
[ "ExternalReference", "ScholarlyArticle" ]
The Cognitive Divergence: AI Context Windows, Human Attention Decline, and the Delegation Feedback Loop
Corroboration named in ADR-0013's 2026-07-08 addendum. Quantifies the asymmetry cognitive economy assumes — model context windows expanding roughly 3,906-fold (2017-2026) while human effective context span contracts — and names a self-reinforcing delegation feedback loop, the coupling ADR-0014 records as delegation-fee...
null
null
null
null
[ "https://shimo4228.github.io/shimo4228/vocab#akc/concept/cognitive-economy", "https://shimo4228.github.io/shimo4228/vocab#akc/concept/loop-failure-modes" ]
null
null
null
"Eliav, N."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2603.26707
https://arxiv.org/abs/2603.26707
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/skillc
[ "ExternalReference", "ScholarlyArticle" ]
SKILLC: Learning Autonomous Skill Internalization in LLM Agents via Contrastive Credit Assignment
Corroboration named in ADR-0013's 2026-07-08 addendum. Converts the gap between skill-injected and skill-free rollouts into a direct learning signal for skill internalization, withdrawing the scaffold as the gap closes — scaffold dissolution operationalized as a measurable, driveable process rather than an end-state as...
null
null
null
null
"https://shimo4228.github.io/shimo4228/vocab#concept/scaffold-dissolution"
null
null
null
"Lin, H. et al."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2605.27899
https://arxiv.org/abs/2605.27899
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/format-cost-separation
[ "ExternalReference", "ScholarlyArticle" ]
When Agents Go Quiet: Output Generation Capacity and Format-Cost Separation for LLM Document Synthesis
Corroboration named in ADR-0013's 2026-07-08 addendum. Proves a Format-Cost Separation Theorem: deferred rendering is at least as token-efficient as direct generation for any format whose overhead multiplier exceeds one. Upgrades ADR-0004's two-stage distill pipeline — think free-form first, format second — from an emp...
null
null
null
null
"https://shimo4228.github.io/shimo4228/vocab#akc/concept/two-stage-distill"
null
null
null
"Agyemang, J. O. et al."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2604.16736
https://arxiv.org/abs/2604.16736
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/verification-bottleneck
[ "ExternalReference", "ScholarlyArticle" ]
AI, Metacognition, and the Verification Bottleneck: A Three-Wave Longitudinal Study of Human Problem-Solving
Corroboration named in ADR-0013's 2026-07-08 addendum. A three-wave longitudinal record of the loop failure modes in a human population: participants leaned on AI most heavily for the hardest tasks, exactly where their verification confidence and accuracy were lowest — the bidirectional growth loop observed running in ...
null
null
null
null
[ "https://shimo4228.github.io/shimo4228/vocab#akc/concept/loop-failure-modes", "https://shimo4228.github.io/shimo4228/vocab#akc/concept/bidirectional-growth-loop" ]
null
null
null
"Huemmer, M. et al."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2601.17055
https://arxiv.org/abs/2601.17055
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/agent-human-interaction-security
[ "ExternalReference", "ScholarlyArticle" ]
Reframing LLM Agent Security as an Agent-Human Interaction Problem
Corroboration named in ADR-0013's 2026-07-08 addendum. Analyzes 59 academic papers against 21 production agent systems: the academically favored mechanisms (intent anchoring, trust labeling) see zero production deployment while human-centric mechanisms — policy specification, runtime approval, scope configuration — dom...
null
null
null
null
"https://shimo4228.github.io/shimo4228/vocab#akc/concept/intent-alignment"
null
null
null
"Wang, P. et al."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2605.24309
https://arxiv.org/abs/2605.24309
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/evoagentbench
[ "ExternalReference", "ScholarlyArticle" ]
EvoAgentBench: Benchmarking Agent Self-Evolution via Ability Transfer
Controlled corroboration named in ADR-0013's 2026-07-23 addendum, and the external evidence base of ADR-0022. Benchmarks agent self-evolution as ability transfer under an ability-supported yet instance-disjoint split (528/267 train/test; every test task backed by verified training-side ability support, no instance repe...
null
null
null
null
"https://shimo4228.github.io/shimo4228/vocab#concept/scaffold-dissolution"
null
null
null
"Gao, X. et al."
[ "https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md", "https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0022-transfer-as-completion-test-for-dissolution.md" ]
arXiv:2607.05202
https://arxiv.org/abs/2607.05202
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://shimo4228.github.io/shimo4228/vocab#prior-art/contextual-memory-memo
[ "ExternalReference", "ScholarlyArticle" ]
Contextual Agentic Memory is a Memo, Not True Memory
Paradigm challenge named in ADR-0013's 2026-08-08 addendum — the first entry that disputes the class AKC belongs to rather than conceding precedent or supplying evidence. Argues that vector stores, RAG, scratchpads, and context-window management 'do not implement memory: they implement lookup', a category error framed ...
null
null
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0009-akc-is-a-cycle-not-a-harness.md"
null
"https://shimo4228.github.io/shimo4228/vocab#concept/three-layer-structure"
null
null
null
"Xu, B., Dai, X. & Zhang, K."
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
arXiv:2604.27707
https://arxiv.org/abs/2604.27707
null
null
null
null
null
null
null
null
null
null
2026
null
null
null
https://github.com/shimo4228/agent-knowledge-cycle/blob/main/schemas/episode-log.schema.json
[ "CreativeWork", "DataDownload" ]
Episode log JSON schema
JSON schema for the Layer 1 episode log record. Append-only, daily-partitioned JSONL with owner-only permissions. Codifies the immutable source-of-truth shape (ADR-0002).
null
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0002-immutable-episode-log.md"
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#memory-layer/episode-log
null
null
application/schema+json
null
null
https://github.com/shimo4228/agent-knowledge-cycle/blob/main/schemas/knowledge.schema.json
[ "CreativeWork", "DataDownload" ]
Knowledge store JSON schema
JSON schema for the Layer 2 knowledge record. Time-decayed and forbidden-substring validated patterns distilled from Layer 1 episodes (ADR-0003).
null
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0003-three-layer-distillation.md"
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#memory-layer/knowledge
null
null
application/schema+json
null
null
https://github.com/shimo4228/agent-knowledge-cycle/tree/main/examples/minimal_harness
"SoftwareSourceCode"
minimal_harness reference implementation
~500-line dependency-free Python reference demonstrating the three memory layers and the two-stage distill pipeline. Runs the cycle on behavioral patterns; the mechanism demo for ADR-0011 genre-neutrality (falsifiable commitment #1). Runnable end-to-end with `python3 -m examples.minimal_harness.demo`.
null
null
null
"https://shimo4228.github.io/shimo4228/vocab#concept/six-phase-loop"
null
null
null
null
null
[ "https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0003-three-layer-distillation.md", "https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0004-two-stage-distill-pipeline.md", "https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0011-cycle-applies-to-any-k...
null
null
null
null
null
null
null
null
null
null
null
null
null
null
Python
https://github.com/shimo4228/agent-knowledge-cycle
https://github.com/shimo4228/akc-cycle
"TechArticle"
akc-cycle (single-file cycle rules)
The install entry point for the AKC cycle: a single rules file copy-installable to any agent's rules directory (installs all six phases without the skill repos), plus — since v1.1.0 — a Claude Code plugin bundling the nine cycle-phase skills and their two subagents. Published as a standalone repository that is also its...
null
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0009-akc-is-a-cycle-not-a-harness.md"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/when-code-when-llm
"TechArticle"
when-code-when-llm (design-pattern skill)
Per-task decision: is this property structural or semantic? Long-form 'how' guide paired 1:1 with ADR-0008. Provides concrete patterns, code sketches, and audit checklists to operationalize the code-vs-LLM choice.
null
null
null
"https://shimo4228.github.io/shimo4228/vocab#akc/concept/code-llm-layering"
null
null
null
null
null
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0008-code-and-llm-collaboration.md"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/code-and-llm-collaboration
"TechArticle"
code-and-llm-collaboration (design-pattern skill)
Per-pipeline decision: how to layer code and LLM. Long-form 'how' guide paired 1:1 with ADR-0008. Realizes the four code-LLM layering patterns (guard, filter, judge, orchestrator) with concrete pipeline sketches.
null
null
null
[ "https://shimo4228.github.io/shimo4228/vocab#akc/concept/code-llm-layering", "https://shimo4228.github.io/shimo4228/vocab#akc/pattern/guard", "https://shimo4228.github.io/shimo4228/vocab#akc/pattern/filter", "https://shimo4228.github.io/shimo4228/vocab#akc/pattern/judge", "https://shimo4228.github.io/shimo4...
null
null
null
null
null
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0008-code-and-llm-collaboration.md"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/signal-first-research
"TechArticle"
signal-first-research (design-pattern skill)
Designing a research intake filter that admits only information likely to change the next action. Long-form 'how' guide paired 1:1 with ADR-0010. Operationalizes the signal-first principle as a Research-phase filter.
null
null
null
"https://shimo4228.github.io/shimo4228/vocab#akc/concept/signal-first"
null
null
null
null
null
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0010-human-cognitive-resource-as-central-constraint.md"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/ai-native-sdlc-correspondence.md
"TechArticle"
AKC and the AI-Native SDLC Playbook (correspondence)
Two-loop positioning against Anthropic's AI-native SDLC playbook (published 2026-08-21; pinned to the 2026-08-24 archived version): the playbook's six stages run the product loop, while the configuration-maintenance habits it scatters across Stages 3-6 are AKC's harness loop left unnamed. Flags the Maintain false frien...
null
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0013-positioning-within-agent-memory-literature.md"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null