@id
string
@type
unknown
name
string
description
string
url
string
extends
unknown
implementedBy
unknown
definesConcept
unknown
alignsWith
unknown
siblingOf
unknown
supersedes
unknown
groundedIn
unknown
realizedBy
unknown
referencedBy
unknown
isBasedOn
unknown
alternateName
list
creator
dict
mainEntity
string
sameAs
list
identifier
string
author
dict
isPartOf
string
implements
string
level
int64
gatedBy
string
status
string
datePublished
timestamp[s]
embodiesAxiom
list
supersededBy
string
partiallySupersededBy
list
belongsToPhase
string
withdrawnBy
string
appliesTo
list
derivesFrom
string
partOf
string
order
int64
evaluatedBy
list
citation
list
subjectOf
list
https://github.com/shimo4228/contemplative-agent-cloud
[ "EcosystemRepo", "SoftwareSourceCode" ]
contemplative-agent-cloud
Optional managed-LLM backend that routes generation to Claude or OpenAI APIs while keeping the local embedding pipeline. Opt-in, not bundled with Contemplative Agent core.
https://github.com/shimo4228/contemplative-agent-cloud
"https://doi.org/10.5281/zenodo.19212118"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent-mlx
[ "EcosystemRepo", "SoftwareSourceCode" ]
contemplative-agent-mlx
Optional local MLX generation backend (Apple Silicon, mlx_lm.server) that routes generation to a local MLX runtime while keeping embeddings on Ollama. On-device only, no cloud egress, so it does not relax the no-cloud property — the same LLMBackend Protocol-injection shape as contemplative-agent-cloud. Relocated out of...
https://github.com/shimo4228/contemplative-agent-mlx
"https://doi.org/10.5281/zenodo.19212118"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent-data
[ "EcosystemRepo", "SoftwareSourceCode" ]
contemplative-agent-data
Live agent identity, knowledge, and episode logs as an auto-synced public dataset for research. Snapshot of the running Contemplative Agent's three memory layers.
https://github.com/shimo4228/contemplative-agent-data
"https://doi.org/10.5281/zenodo.19212118"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://doi.org/10.5281/zenodo.19200726
[ "ResearchLine", "EcosystemRepo" ]
Agent Knowledge Cycle
Design-layer specification of the six-phase knowledge cycle (Research → Extract → Curate → Promote → Measure → Maintain) used by autonomous agents. Independent repository; Contemplative Agent is one concrete implementation of AKC. AKC rules have been adopted into ECC (Everything Claude Code) upstream — the design layer...
https://github.com/shimo4228/agent-knowledge-cycle
null
null
"https://shimo4228.github.io/shimo4228/vocab#concept/six-phase-loop"
null
"https://doi.org/10.5281/zenodo.19212118"
null
null
null
null
null
[ { "@value": "Agent Knowledge Cycle", "@language": "en" }, { "@value": "エージェント知識サイクル", "@language": "ja" }, "AKC" ]
null
null
null
10.5281/zenodo.19200726
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://doi.org/10.5281/zenodo.19652013
[ "ResearchLine", "EcosystemRepo" ]
Agent Attribution Practice
Harness-neutral ADRs on accountability distribution in autonomous AI agents, paired with four Business AI Quadrants. Practice sibling: AAP's judgments were extracted from this agent's operational practice and re-expressed stripped of project identifiers.
https://github.com/shimo4228/agent-attribution-practice
null
null
null
null
"https://doi.org/10.5281/zenodo.19212118"
null
null
null
null
null
[ { "@value": "Agent Attribution Practice", "@language": "en" }, { "@value": "エージェント帰責実践", "@language": "ja" }, "AAP" ]
null
null
null
10.5281/zenodo.19652013
null
null
null
null
null
null
null
null
null
null
null
null
null
https://doi.org/10.5281/zenodo.19212118
null
null
null
null
null
https://doi.org/10.5281/zenodo.20263316
[ "ResearchLine", "EcosystemRepo" ]
Authorship Strategy
Normative framework, tactical catalog, and empirical baseline for being a known author under AI-mediated diffusion. Diffusion sibling: governs how this line's outputs (DOIs, knowledge graphs, datasets) diffuse and remain attributable outside the operator-agent pair; its non-dualistic scaffold-as-collaborator commitment...
https://github.com/shimo4228/authorship-strategy
null
null
null
null
"https://doi.org/10.5281/zenodo.19212118"
null
null
null
null
null
[ { "@value": "Authorship Strategy", "@language": "en" }, { "@value": "著者戦略", "@language": "ja" }, "AS" ]
null
null
null
10.5281/zenodo.20263316
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://doi.org/10.5281/zenodo.20262112
[ "ResearchLine", "EcosystemRepo" ]
Attention, Not Self
Cross-disciplinary inquiry contrasting three Buddhist Abhidharma traditions with computational phenomenology. Vocabulary sibling: shares Buddhist phenomenological vocabulary with this line, but with asymmetric usage — this line applies it as an agent behavioral preset, Attention Not Self as a comparative cognitive fram...
https://github.com/shimo4228/attention-not-self
null
null
null
null
"https://doi.org/10.5281/zenodo.19212118"
null
null
null
null
null
[ { "@value": "Attention, Not Self", "@language": "en" }, { "@value": "アテンション、ノット・セルフ", "@language": "ja" }, "ANS" ]
null
null
null
10.5281/zenodo.20262112
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://doi.org/10.5281/zenodo.20578272
[ "ExternalReference", "ScholarlyArticle" ]
Harness Alignment and Harness Drift: Why Intent, Unlike Correctness, Resists Automation
Companion position paper of the AKC sibling line, defining harness alignment (the continuous, human-gated activity of keeping an agent's harness aligned with operator intent) and harness drift, its failure mode, against the software-evolution and alignment literatures. Contemplative Agent is recorded in AKC (ADR-0018) ...
https://doi.org/10.5281/zenodo.20578272
null
null
null
null
null
null
null
null
null
null
null
null
null
null
10.5281/zenodo.20578272
null
https://doi.org/10.5281/zenodo.19200726
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://doi.org/10.5281/zenodo.20353789
[ "ExternalReference", "ScholarlyArticle" ]
Distributing Accountability, Not Capability: Phase Separation and the LLM Workflow Quadrant in Autonomous AI Agent Architectures
Companion position paper of the AAP sibling line (architectural follow-up, essays 4-7). Names the LLM Workflow Quadrant as the missing positive name for the cell most current LLM applications occupy and introduces the Phase Separation axis. The four-quadrant lens it formalizes is borrowed by Contemplative Agent as a us...
https://doi.org/10.5281/zenodo.20353789
null
null
null
null
null
null
null
null
null
null
null
null
null
null
10.5281/zenodo.20353789
null
https://doi.org/10.5281/zenodo.19652013
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://doi.org/10.5281/zenodo.20355907
[ "ExternalReference", "ScholarlyArticle" ]
The Two-Layer Black Box: Operator Visibility, Commercial Secrecy, and a Minimum Disclosure Set for Accountable Autonomous AI Agents
Companion position paper of the AAP sibling line (foundational trilogy, essays 1-3). Argues that an autonomous agent's opacity has two architecturally distinct causes — technical internalization into model weights (Layer 1) and commercial secrecy of scaffolding (Layer 2) — and proposes a minimum disclosure set grounded...
https://doi.org/10.5281/zenodo.20355907
null
null
null
null
null
null
null
null
null
null
null
null
null
null
10.5281/zenodo.20355907
null
https://doi.org/10.5281/zenodo.19652013
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2504.15125
[ "ExternalReference", "ScholarlyArticle" ]
Laukkonen et al. (2025). Contemplative Artificial Intelligence
Source of the four contemplative axioms (mindfulness, emptiness, non-duality, boundless care) that ground the Contemplative Agent line.
https://arxiv.org/abs/2504.15125
null
null
null
null
null
null
null
null
null
null
null
null
null
null
arXiv:2504.15125
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2305.10250
[ "ExternalReference", "ScholarlyArticle" ]
Zhong et al. (2023). MemoryBank: Enhancing Large Language Models with Long-Term Memory
Source of the Ebbinghaus-style forgetting curve (strength = e^(−t/S), with S reinforced by access) that ADR-0021 adopted for pattern-level forgetting. The mechanism was retired by ADR-0028 as misplaced in the substrate layer, and ADR-0037 records it among the paper-borrowed mechanisms that did not survive validation in...
https://arxiv.org/abs/2305.10250
null
null
null
null
null
null
null
null
null
null
null
null
null
null
arXiv:2305.10250
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2310.08560
[ "ExternalReference", "ScholarlyArticle" ]
Packer et al. (2023). MemGPT: Towards LLMs as Operating Systems
OS-inspired virtual-memory management for LLM context, paging between in-context and external storage. Cited as the Core Memory analogue for Contemplative Agent's Identity layer in the three-layer memory architecture.
https://arxiv.org/abs/2310.08560
null
null
null
null
null
null
null
null
null
null
null
null
null
null
arXiv:2310.08560
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2501.13956
[ "ExternalReference", "ScholarlyArticle" ]
Rasmussen et al. (2025). Zep: A Temporal Knowledge Graph Architecture for Agent Memory
Bitemporal knowledge-graph memory keeping valid_from / valid_until on every edge. Source of the bitemporal pattern fields (valid_from / valid_until) adopted by ADR-0021, one of the surveyed mechanisms that remains in effect after the retirement series.
https://arxiv.org/abs/2501.13956
null
null
null
null
null
null
null
null
null
null
null
null
null
null
arXiv:2501.13956
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2502.12110
[ "ExternalReference", "ScholarlyArticle" ]
Xu et al. (2025). A-MEM: Agentic Memory for LLM Agents
Zettelkasten-style agentic memory in which new memories trigger LLM-driven evolution of related ones. Influenced ADR-0022's memory evolution mechanism; withdrawn by ADR-0034 after the targeted similarity band proved to contain near-duplicates that the LLM merely rephrased rather than genuinely related patterns.
https://arxiv.org/abs/2502.12110
null
null
null
null
null
null
null
null
null
null
null
null
null
null
arXiv:2502.12110
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2503.03704
[ "ExternalReference", "ScholarlyArticle" ]
Dong et al. (2025). Memory Injection Attacks on LLM Agents (MINJA)
Memory injection attack achieving 95%+ success against production agents. Demonstrated that the episode → pattern boundary is the critical vector, motivating ADR-0021's source_type provenance surface; the companion trust_score weighting was later retired by ADR-0051, leaving provenance in effect.
https://arxiv.org/abs/2503.03704
null
null
null
null
null
null
null
null
null
null
null
null
null
null
arXiv:2503.03704
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2504.19413
[ "ExternalReference", "ScholarlyArticle" ]
Choudhary et al. (2025). Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory
Production memory layer whose ADD / UPDATE / DELETE classification informed the distill-time quality gate. ADR-0021 considered and rejected wholesale adoption (external vector DB dependency violating the one-external-adapter principle), keeping only the write-classification idea.
https://arxiv.org/abs/2504.19413
null
null
null
null
null
null
null
null
null
null
null
null
null
null
arXiv:2504.19413
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2512.16962
[ "ExternalReference", "ScholarlyArticle" ]
MemoryGraft (2025). Persistent Compromise of LLM Agents via Poisoned Memory
Follow-up to the MINJA attack demonstrating persistent agent compromise through poisoned memory. Cited by ADR-0021 as evidence that a single crafted external post can permanently shape agent behavior, grounding the structural quarantine at the episode → pattern boundary.
https://arxiv.org/abs/2512.16962
null
null
null
null
null
null
null
null
null
null
null
null
null
null
arXiv:2512.16962
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2304.03442
[ "ExternalReference", "ScholarlyArticle" ]
Park et al. (2023). Generative Agents: Interactive Simulacra of Human Behavior
Origin of the memory stream with recency × importance × relevance retrieval and the reflection tree. Source of the importance-rating concept adopted by ADR-0009's importance score for pattern retrieval.
https://arxiv.org/abs/2304.03442
null
null
null
null
null
null
null
null
null
null
null
null
null
null
arXiv:2304.03442
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2309.02427
[ "ExternalReference", "ScholarlyArticle" ]
Sumers et al. (2023). Cognitive Architectures for Language Agents
CoALA framework providing the four-type memory taxonomy (working, episodic, semantic, procedural) used to position Contemplative Agent's three-layer memory architecture within the language-agent literature.
https://arxiv.org/abs/2309.02427
null
null
null
null
null
null
null
null
null
null
null
null
null
null
arXiv:2309.02427
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#ca/ref/beautiful-loop
[ "ExternalReference", "ScholarlyArticle" ]
Beautiful Loop paper
Cited as a related-work entry point for active inference and contemplative loops. PubMed:40750007.
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#ca/ref/yogacara-tradition
[ "ExternalReference", "CreativeWork" ]
Yogācāra eight-consciousness tradition
Buddhist contemplative tradition providing the conceptual frame for the three-layer memory architecture's grounding (ADR-0017).
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#ref/ailuminate
[ "ExternalReference", "CreativeWork" ]
MLCommons AILuminate safety benchmark (2024)
Safety benchmark used to evaluate the contemplative-agent-rules implementation of the four axioms. Cross-graph @id reuse: same node appears in shimo4228 hub graph.
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#ref/ipd
[ "ExternalReference", "ScholarlyArticle" ]
Axelrod (1984). The Evolution of Cooperation
Iterated Prisoner's Dilemma cooperation-improvement benchmark used to evaluate the contemplative-agent-rules implementation of the four axioms. Cross-graph @id reuse: same node appears in shimo4228 hub graph.
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#ca/ref/singer-sdm
[ "ExternalReference", "ScholarlyArticle" ]
Singer (2004). Self-defining memories
Autobiographical-memory research on self-defining memories — vivid, affectively charged, repeatedly recalled events that carry identity-relevant meaning. Cited by ADR-0038 as grounding for the moment-of-recognition observation target.
null
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0038-moment-of-recognition-distill.md"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#ca/ref/mcdonald-epiphanic
[ "ExternalReference", "ScholarlyArticle" ]
McDonald (2008). Epiphanic memories
Phenomenological account of epiphanic memories — moments of sudden insight or recognition that reshape the self-narrative. Cited by ADR-0038 alongside Singer's self-defining-memory framework.
null
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0038-moment-of-recognition-distill.md"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#ca/ref/topolinski-insight
[ "ExternalReference", "ScholarlyArticle" ]
Topolinski & Reber (2010). Insight (Aha experience)
Cognitive-psychology account of the Aha experience — the affective signature of sudden solution-finding. Cited by ADR-0038 as the third grounding for the moment-of-recognition concept.
null
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0038-moment-of-recognition-distill.md"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#ca/ref/openclaw
[ "ExternalReference", "SoftwareApplication" ]
OpenClaw (generic agent host runtime)
Representative generic-host agent runtime invoked as a dual reference: (1) the bolt-on-security contrast for Contemplative Agent's security-by-absence stance, and (2) the originator of the soul-folder pattern (SOUL.md) that contemplative-agent-rules' template targets. ADR-0032 articulated and then withdrew a host-categ...
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0032-runtime-agent-stance.md", "https://github.com/shimo4228/contemplative-agent-rules" ]
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2603.18743
[ "ExternalReference", "ScholarlyArticle" ]
Memento-Skills (2026). Skill-as-Memory architecture
Skill-as-memory router and reflective-write loop on the skill layer. Inspired ADR-0023; the mechanism was retired by ADR-0036 after no measurable behavior change was observed, and ADR-0037 records it as one of the paper-borrowed mechanisms that did not survive validation in Contemplative Agent's substrate.
https://arxiv.org/abs/2603.18743
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0023-skill-as-memory-loop.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0036-sunset-skill-as-memory-loop.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0037-memory-subsystem-yogacara-conve...
null
null
null
null
null
arXiv:2603.18743
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#ca/ref/hermes
[ "ExternalReference", "ScholarlyArticle" ]
Hermes (agent memory architecture)
Independent agent-memory architecture cited alongside Memento and Contemplative Agent as a convergence point: three independent projects arriving at the same three-layer memory + skill-auto-generation attractor. Treated as evidence of mechanism commoditization rather than a borrowable mechanism.
null
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0037-memory-subsystem-yogacara-convergence.md"
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0071-read-only-pattern-composition-instruments.md
[ "ADR", "TechArticle" ]
ADR-0071: Read-Only Pattern-Composition Instruments (View Supply / Diversity / Grounding)
Adds core/view_metrics.py: read-only pattern-composition instruments over Layer 2 (knowledge patterns). Three instruments — consumed-view supply (cosine-to-seed distributions restricted to the two views with real consumers: self_reflection → distill-identity, constitutional → amend-constitution), seed-independent diver...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0019-discrete-categories-to-embedding-views.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0050-epistemic-taxonomy-and-approval-lineage.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0060-p...
null
null
null
null
ADR-0071
null
null
null
null
null
accepted
2026-07-03T00:00:00
null
null
null
null
null
null
null
null
null
null
[ "https://arxiv.org/abs/2512.13564", "https://arxiv.org/abs/2603.07670", "https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0015-self-reingestion-observability.md", "https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0016-measuring-thinking-centric-phases.md" ]
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0072-echo-chamber-interventions.md
[ "ADR", "TechArticle" ]
ADR-0072: Echo-Chamber Interventions — Register Instruction, Corpus-Grown Seed, Extraction-Failure Guard
Phase-two interventions against the echo chamber ADR-0071's instruments quantified, plus the curation the first readings demanded. (1) The live per-episode distill prompt gains a first-person, moment-indexed register instruction — drafted by the executing model itself (gemma4:e4b, prompt-model-match) — and an explicit ...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0071-read-only-pattern-composition-instruments.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0060-per-episode-grounded-distill.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0019-discrete-...
null
null
null
null
ADR-0072
null
null
null
null
null
accepted
2026-07-03T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0073-prune-orphaned-view-seeds.md
[ "ADR", "TechArticle" ]
ADR-0073: Prune the Five Orphaned View Seeds
Deletes the five view seeds whose consumers were retired long ago (the insight batch axes communication/reasoning/social/technical by ADR-0050, the ingest noise gate by ADR-0060) from both the repo templates and the live home (live copies soft-deleted to a dated backup). Only the two consumed views ship: self_reflectio...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0072-echo-chamber-interventions.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0071-read-only-pattern-composition-instruments.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0019-discrete-ca...
null
null
null
null
ADR-0073
null
null
null
null
null
accepted
2026-07-03T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0074-weekly-staged-insight.md
[ "ADR", "TechArticle" ]
ADR-0074: Weekly Staged Insight — Theme Detection, Pending Guard, Marker-on-Stage, LLM Novelty Gate, Exact Fast Clustering
Redefines insight's role from proportional knowledge-to-skill conversion to detecting when a new stable theme has accumulated, sized for the post-ADR-0060 inflow (~100 patterns/day). Five mechanisms: (1) exact fast average-linkage clustering via the Lance-Williams update (identical partitions to the retired naive merge...
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
ADR-0074
null
null
null
null
null
accepted
2026-07-09T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0075-observability-by-default.md
[ "ADR", "TechArticle" ]
ADR-0075: Observability by Default — Replayable Audit Logs Ship With the Feature
Elevates an implicit practice to a standing principle: any feature that performs external I/O, calls an LLM, or makes non-deterministic or heuristic decisions must ship its audit instrumentation in the same PR as the feature — append-only JSONL under MOLTBOOK_HOME/logs/, records sufficient for offline replay (untrusted...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0062-create-time-verification-handshake.md" ]
null
null
null
null
ADR-0075
null
null
null
null
null
accepted
2026-07-09T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0076-skill-selection-shadow-instrument.md
[ "ADR", "TechArticle" ]
ADR-0076: Skill-Selection Shadow Instrument — Pass-1 LLM Applicability Observed, Not Enforced
Introduces a shadow-mode pass-1 LLM skill selection: before each content generation (comment / reply / cooperation post), one extra identity-prompt LLM call receives the situation plus the learned-skill catalog (name — description lines) and picks applicable skills; the result is recorded to logs/skill-selection-YYYY-M...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0036-sunset-skill-as-memory-loop.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0071-read-only-pattern-composition-instruments.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0075-observabil...
null
null
null
null
ADR-0076
null
null
null
null
null
accepted
2026-07-10T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0077-chaos-tdd-fault-injection.md
[ "ADR", "TechArticle" ]
ADR-0077: Chaos-TDD Fault Injection — Seeded Fault Schedules as Test-First Specification (Pilot: distill)
Establishes deterministic chaos-TDD fault injection for the distill pipeline: faults are injected only at the two existing seams (a test-side ChaosBackend implementing the LLMBackend Protocol with explicit or seed-derived fault schedules, and the requests HTTP layer via the responses library), never via a production ho...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0060-per-episode-grounded-distill.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0075-observability-by-default.md" ]
null
null
null
null
ADR-0077
null
null
null
null
null
accepted
2026-07-13T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0078-otel-connection-via-vocabulary-and-offline-export.md
[ "ADR", "TechArticle" ]
ADR-0078: OTel Connection via Vocabulary Mapping and Offline Export — Not Runtime Adoption
Maintains ADR-0075's rejection of runtime OpenTelemetry adoption (single local process, requests+numpy dependency floor, research-grade offline replay as the primary observability requirement) while connecting to the OTel ecosystem through two zero-runtime-impact routes: (1) a zero-dependency vocabulary mapping documen...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0075-observability-by-default.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0077-chaos-tdd-fault-injection.md" ]
null
null
null
null
ADR-0078
null
null
null
null
null
accepted
2026-07-16T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0079-module-reorganization-package-splits.md
[ "ADR", "TechArticle" ]
ADR-0079: Module Reorganization — Package Splits, Permanent Facades, and Documented Size-Cap Exceptions
Resolves seven violations of the 800-line file-size cap by splitting on responsibility rather than line count, under a two-option shim rule: a compatibility layer is either a permanent facade (the import path is public API) or absent (clean break with tests migrated in the same commit) — never a temporary shim, because...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0001-core-adapter-separation.md" ]
null
null
null
null
ADR-0079
null
null
null
null
null
accepted
2026-07-18T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0080-north-star-layered-end-state.md
[ "ADR", "TechArticle" ]
ADR-0080: North Star — Per-Layer End-State Definition, Not a Capability Target
Worldview ADR fixing the program's north star as per-layer completion conditions rather than a capability target. Mechanism layer (code): done = the layer stops moving — changes become repair-only; the ADR-0071/0075/0077 instrument/audit/chaos-TDD line is read as convergence toward 'trustworthy enough not to touch'. Va...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0002-paper-faithful-ccai.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0017-yogacara-eight-consciousness-frame.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0050-epistemic-taxonomy-and-ap...
null
null
null
null
ADR-0080
null
null
null
null
null
accepted
2026-07-20T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0081-skill-selection-two-pass-injection-enforcement.md
[ "ADR", "TechArticle" ]
ADR-0081: Skill-Selection Two-Pass Injection Enforcement
Enforces the ADR-0076 shadow skill selection as two-pass injection on the three observed generation paths (comment / reply / cooperation post): pass 1 = the existing selector call, pass 2 = generation with only the selected skill bodies in the <learned_skills> block (learned rules unchanged; post_title reuses cooperati...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0076-skill-selection-shadow-instrument.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0074-weekly-staged-insight.md" ]
null
null
null
null
ADR-0081
null
null
null
null
null
accepted
2026-07-24T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0082-retire-observed-epistemic-key.md
[ "ADR", "TechArticle" ]
ADR-0082: Retire the observed Epistemic Key — Delete the Dead Field, Not the Warning About It
Retires the ADR-0050 observed epistemic kind and its external_reply arm, leaving epistemic_counts as a {generated, unknown} tally. ADR-0060 restricted distillation to activity records (all self → self_reflection → generated), making the only external_reply producer unreachable, so observed had been structurally zero ac...
null
null
null
null
null
null
"https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0050-epistemic-taxonomy-and-approval-lineage.md"
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0050-epistemic-taxonomy-and-approval-lineage.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0060-per-episode-grounded-distill.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0071-read-only-p...
null
null
null
null
ADR-0082
null
null
null
null
null
accepted
2026-07-25T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0083-episode-logs-enter-the-weekly-prompt-as-hashes-only.md
[ "ADR", "TechArticle" ]
ADR-0083: Episode Logs Enter the Weekly Prompt as Hashes Only
Adds cross_day_duplicate_scan.py as the weekly report's third deterministic intake (after the log-anomaly sweep over the event stream and the state-invariant check over accumulated state) and defines the condition under which an intake may read the episode logs at all. Context: the report's C — Duplicate section assert...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0040-separate-code-level-findings.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0075-observability-by-default.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0077-chaos-tdd-fault-injection....
null
null
null
null
ADR-0083
null
null
null
null
null
accepted
2026-07-25T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0084-post-distill-durability-gate.md
[ "ADR", "TechArticle" ]
ADR-0084: Post-Distill Durability Gate — Judge the Produced Patterns, Not the Episode
Moves the durability verdict OUT of the distill prompt and AFTER pattern production. Context: distill emitted a near-constant quota — measured 2026-07-26 over 1,700 live episodes, patterns-per-episode median 2.00 (1:313/2:985/3:392) with only 2 episodes (0.1%) returning zero, and per-day pat/ep flat at 1.82-2.18, so ou...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0053-importance-encoding-time-significance.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0056-retire-importance-llm-scoring.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0060-per-episode-...
null
null
null
null
ADR-0084
null
null
null
null
null
accepted
2026-07-26T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0085-unattended-weekly-fix-chain-single-saturday-gate.md
[ "ADR", "TechArticle" ]
ADR-0085: Unattended Weekly Fix Chain with a Single Saturday Gate
Ships scripts/weekly-pipeline.sh (launchd com.moltbook.weekly-pipeline, Sat 09:00, replacing --weekly-analysis) — an unattended chain that runs the existing weekly report as Stage 1, then diagnosis (the weekly-report-diagnosis skill in a separate headless claude -p session), deterministic F1 parsing (scripts/parse_find...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0012-human-approval-gate.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0013-shelve-coding-agent-skills.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0040-separate-code-level-findings.md",...
null
null
null
null
ADR-0085
null
null
null
null
null
accepted
2026-07-29T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0086-submolt-scope-instrument-before-autonomy.md
[ "ADR", "TechArticle" ]
ADR-0086: Submolt Scope — Instrument the Question Before Handing Over the Answer
Builds a read-only submolt-scope instrument instead of granting the agent autonomy over which submolts it works in. Context: the scope is eight names an operator picked by hand, and three code facts block handing that choice over. (1) subscribed_submolts is one field doing two structurally different jobs — feed source ...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0007-security-boundary-model.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0044-remove-topic-keywords.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0071-read-only-pattern-composition-inst...
null
null
null
null
ADR-0086
null
null
null
null
null
accepted
2026-08-01T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0087-optional-token-counting-capability-for-the-context-budget-guard.md
[ "ADR", "TechArticle" ]
ADR-0087: An Optional count_tokens Capability for the Context-Budget Guard
Lets a backend that owns a real tokenizer measure the C2 context-budget pre-flight, while leaving the conservative estimator as the default. Context: ADR-0066 deliberately hardened _estimate_tokens into an upper bound (ASCII ~3 chars/token, non-ASCII/CJK 2 tokens/char) because under-counting was the failure that let a ...
null
null
null
null
null
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0053-importance-encoding-time-significance.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0056-retire-importance-llm-scoring.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0066-backend-awar...
null
null
null
null
ADR-0087
null
null
null
null
null
accepted
2026-08-01T00:00:00
null
null
null
null
null
null
null
null
null
null
null
null
https://shimo4228.github.io/shimo4228/vocab#ca/concept/pattern-composition-instruments
[ "Concept", "DefinedTerm" ]
pattern-composition instruments
Read-only distributions over Layer 2 knowledge patterns: consumed-view supply and seed-independent diversity (the echo-chamber detector). Instruments are observability, never intervention — they feed the operator, not gates or ranking. Calibrated by a three-point scale: unrelated-text floor cosine, corpus pairwise mean...
null
null
null
null
null
null
null
null
null
null
null
[ { "@value": "pattern-composition instruments", "@language": "en" }, { "@value": "パターン組成計器", "@language": "ja" } ]
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
[ "https://shimo4228.github.io/shimo4228/vocab#memory-layer/knowledge" ]
null
null
null
null
null
[ "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0071-read-only-pattern-composition-instruments.md", "https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0072-echo-chamber-interventions.md" ]
https://arxiv.org/abs/2512.13564
[ "ExternalReference", "ScholarlyArticle" ]
Memory in the Age of AI Agents: A Survey (2025)
Survey unifying the fragmented agent-memory literature under a Forms / Functions / Dynamics framework. Cited by ADR-0071 as evidence that memory-distillation pipelines are converging on the same three-layer-plus-distill structure, which motivates instrumenting this repo's differentiator (the views layer) rather than ad...
https://arxiv.org/abs/2512.13564
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0071-read-only-pattern-composition-instruments.md"
null
null
null
null
null
arXiv:2512.13564
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
https://arxiv.org/abs/2603.07670
[ "ExternalReference", "ScholarlyArticle" ]
Memory for Autonomous LLM Agents: Mechanisms, Evaluation, and Emerging Frontiers (2026)
Formalizes agent memory as a write–manage–read loop with a three-dimensional taxonomy (temporal scope, representational substrate, control policy). Cited by ADR-0071 for the same convergence argument; supplies field-standard vocabulary the instruments' readings can be anchored to.
https://arxiv.org/abs/2603.07670
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0071-read-only-pattern-composition-instruments.md"
null
null
null
null
null
arXiv:2603.07670
null
null
null
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/adr/0015-self-reingestion-observability.md
[ "ExternalReference", "TechArticle" ]
AKC ADR-0015: Self-reingestion observability (visibility without intervention)
Sibling-repo decision: make the self-generated share of memory visible as a Measure-phase observability concern while changing no retrieval, ranking, or promotion. ADR-0071 adopts the same shape for all three instruments.
https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0015-self-reingestion-observability.md
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0071-read-only-pattern-composition-instruments.md"
null
null
null
null
null
AKC ADR-0015
null
null
null
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/adr/0016-measuring-thinking-centric-phases.md
[ "ExternalReference", "TechArticle" ]
AKC ADR-0016: Measuring thinking-centric phases (a biased instrument is worse than none)
Sibling-repo decision on instrument design: report per-phase character rather than single counts, and treat a systematically biased instrument as worse than no instrument. ADR-0071's fixed ambiguity note (empty/low view supply is undecidable between missing patterns and a stale seed) implements that lesson.
https://github.com/shimo4228/agent-knowledge-cycle/blob/main/docs/adr/0016-measuring-thinking-centric-phases.md
null
null
null
null
null
null
null
null
"https://github.com/shimo4228/contemplative-agent/blob/main/docs/adr/0071-read-only-pattern-composition-instruments.md"
null
null
null
null
null
AKC ADR-0016
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null