Graph Report - . (2026-05-01)

Corpus Check

448 files · ~249,404 words
Verdict: corpus is large enough that graph structure adds value.

Summary

824 nodes · 1766 edges · 43 communities detected
Extraction: 94% EXTRACTED · 6% INFERRED · 0% AMBIGUOUS · INFERRED: 111 edges (avg confidence: 0.79)
Token cost: 0 input · 0 output

God Nodes (most connected - your core abstractions)

MOC PUMA Master - 79 edges
Master Index — Johnny Decimal - 44 edges
MOC Methods Frameworks - 37 edges
MOC Research Pipeline - 34 edges
EX Hypotheses H1 H2 - 33 edges
MOC Tools Stack - 32 edges
MOC Zettelkasten PUMA - 32 edges
MOC LLM Benchmarks PM AI - 26 edges
PR PUMA Ch3 Methods - 25 edges
Wilcoxon, FINER, Cornell, PRISMA-DFLLM Methods - 24 edges
SP Architecture - 24 edges
SP PUMA Constitution - 23 edges
PUMA Vault Home Dashboard - 22 edges
PR PUMA Ch1 Introduction - 22 edges
RAG, Embeddings, and Vector Databases - 21 edges

Surprising Connections (you probably didn’t know these)

Zettelkasten Knowledge System —semantically_similar_to⇒ LLM Wiki Pattern [INFERRED] [semantically similar] README.md → 20 - Literature/20.7 Blog-Web/LN-Karpathy-2026-LLMWiki.md
CO-STAR Prompt Framework —semantically_similar_to⇒ RCOIF Prompting Framework [INFERRED] [semantically similar] 30 - Permanent/31 Concepts/PN-COSTAR-SelfConsistency.md → index.md
Bounded Autonomy for AI Agents —semantically_similar_to⇒ Wide Safety [INFERRED] [semantically similar] 30 - Permanent/31 Concepts/PN-HITL-BoundedAutonomy.md → 20 - Literature/20.2 Books/LN-Suleiman-2023-ComingWave.md
Context Engineering —semantically_similar_to⇒ Virtual Context Management [INFERRED] [semantically similar] 60 - Resources/65 Bibliographies/BIB-Supplement (Copiar).md → 20 - Literature/20.1 Papers/LLM-Agents-General/LN-Packer-2023-MemGPT.md
Agentic Science Paradigm —semantically_similar_to⇒ Agentic Software Engineering [INFERRED] [semantically similar] 20 - Literature/20.1 Papers/Agentic-Science/LN-Zhang-2025-AgenticScienceSurvey.md → 60 - Resources/65 Bibliographies/BIB-Supplement (Copiar).md
Phase 2 — Literature Discovery & Synthesis —applies⇒ How to Read a Paper (Keshav 2007) [EXTRACTED] 00 - Meta/Assets/research_pipeline.svg → 60 - Resources/65 Bibliographies/BIB-Master-APA7.md
Phase 3 — Hypothesis Calibration & Experimental Design —uses⇒ Wilcoxon Non-parametric Test (1945) [EXTRACTED] 00 - Meta/Assets/research_pipeline.svg → 60 - Resources/65 Bibliographies/BIB-Master-APA7.md

Hyperedges (group relationships)

PUMA Research Pipeline: Datasets + LLMs + Evaluation — concept_tawos_dataset, concept_jira_sr_dataset, ref_llama32, ref_mistral7b, concept_f1_macro, concept_wilcoxon, concept_issue_triage, concept_effort_estimation [INFERRED 0.90]
PUMA Prompting Experiment: CoT + Few-Shot + Zero-Shot evaluated on H1 and H2 — concept_cot, concept_few_shot, concept_costar, concept_self_consistency, concept_h1_triage, concept_h2_estimation, concept_reproducibility [INFERRED 0.85]
PUMA Knowledge Management: GTD + PARA + Zettelkasten + MOC — concept_gtd, concept_para, concept_zettelkasten, concept_moc, concept_johnny_decimal, zk_hub_puma, moc_gtd [INFERRED 0.88]
GTD Review System — [INFERRED 0.85]
Literature Processing Pipeline — [INFERRED 0.85]
Vault Meta Layer — [INFERRED 0.90]
RAG and Vector DB Video Cluster — [INFERRED 0.95]
Research Literature Video Cluster — [INFERRED 0.95]
Dev Tools and IDE Video Cluster — [INFERRED 0.95]
SDD Specs BMAD Video Cluster — [INFERRED 0.90]
PUMA Glossary Ecosystem — [INFERRED 0.90]
Phase 1 Step 4 Conceptual Synthesis Prompt Cluster — [INFERRED 0.85]
Phase 1 Step 5 Hypothesis Generation Prompt Cluster — [INFERRED 0.85]
Research Workflow Support Infrastructure — [INFERRED 0.80]
PUMA Human Oversight Triad: three complementary justifications for human-in-the-loop design — operational (bounded autonomy levels), philosophical (irreducible human knowledge), and governance (wide safety) — concept_hitl, concept_bounded_autonomy, concept_atomic_human [INFERRED]
Systems Business Stack: four business frameworks converging on SmartPMO as an autonomous, documented, self-managing PM system — concept_smart_pmo, concept_franchise_prototype, concept_eos, concept_systems_mindset [INFERRED]
Knowledge Accumulation Paradigms: three related approaches to building persistent, queryable knowledge structures — technical RAG, LLM-maintained wiki, and human Zettelkasten method — concept_rag_pipeline, concept_llm_wiki_pattern, concept_zettelkasten [INFERRED]
PUMA Research Methodology Stack: DSR paradigm + SLR protocol + PRISMA reporting together constitute the complete methodological foundation of PUMA — ln_hevner_2004_dsr_paper, ln_kitchenham_2007_slr_paper, ln_page_2021_prisma2020_paper [EXTRACTED]
AI-as-Scientific-Model Exemplars: Four papers collectively demonstrating that AI can surpass human experts in structured prediction and discovery tasks, used to contextualise PUMA’s PM automation within a broader validated AI trajectory — ln_jumper_2021_alphafold_paper, ln_degrave_2022_plasmacontrol_paper, ln_lam_2023_graphcast_paper, ln_arkanihamed_2026_gluonamplitudes_paper [EXTRACTED]
PUMA’s Three Gap Justifications: Cinkusz motivates systematic prompting comparison, Angermeir motivates reproducibility-first design, and Strubell motivates CodeCarbon integration — together they define PUMA’s three core contributions — ln_cinkusz_2025_cognitiveagentsagilepm_paper, ln_angermeir_2025_reproducibility_paper, ln_strubell_2019_energynlp_paper [INFERRED]
Modular role-based multi-agent patterns collectively informing PUMA Smart PMO architecture — ln_masai_2024_modular_agents, ln_magis_2024_role_team, ln_hong_2023_metagpt_roles, puma_smart_pmo [EXTRACTED]
Alignment risks and governance mechanisms jointly motivating safe agentic AI deployment — ln_haif_2026_hitl, ln_hubinger_2019_deceptive_alignment, ln_authworkflows_2026_auth_model [INFERRED]
Core agentic loop components (reasoning, memory, reflection) composing persistent agent behavior — ln_yao_2022_react_method, ln_park_2023_memory_stream, ln_park_2023_reflection [INFERRED]
Multi-Agent Framework Landscape: concurrent frameworks classified by the Talebirad taxonomy — ln_wu_2023_autogen_autogen, ln_gao_2024_agentscope_agentscope, ln_qian_2023_chatdev_chatdev, ln_talebirad_2023_multiagentsurvey_taxonomy [INFERRED]
Reasoning Enhancement Continuum: from CoT prompting through Reflexion self-critique to self-trained internal rationales — ln_wei_2022_chainofthought_cot, ln_shinn_2023_reflexion_reflexion, ln_zelikman_2024_quietstar_self_training_cot [INFERRED]
Operational AI Evaluation Ecosystem: benchmarks and frameworks for evaluating AI agents in real-world DevOps and incident management tasks — ln_chen_2025_aiopslabs_aiopslabs, ln_incident_management_ai_2023_paper, ln_bruneliere_2022_aidoart_aidoart, ln_liu_2023_agentbench_agentbench [INFERRED]
PUMA Phase 1 Literature Discovery Prompt Pipeline — five sequential prompt templates covering literature exploration, scientific mapping, structured extraction, and PRISMA systematic review — pt_p1s1_007_research_rabbit_prompt, pt_p1s1_004_arxiv_prompt, pt_p1s2_001_connected_papers_prompt, pt_p1s3_003_zotero_curation_prompt, pt_meth_001_prisma_protocol [EXTRACTED]
PUMA Local AI Infrastructure Stack — containerised local LLM inference, embedding, and vector search stack operating without external APIs — glossary_tools_ollama, glossary_tools_qdrant, glossary_tools_llamaindex, glossary_tools_nomic_embed, glossary_tools_docker [EXTRACTED]
PUMA Five-Phase Agentic Research Orchestration — iterative scientific research pipeline from ideation through refinement with feedback loop — research_pipeline_svg_phase1_ideation, research_pipeline_svg_phase2_literature, research_pipeline_svg_phase3_hypothesis, research_pipeline_svg_phase4_review, research_pipeline_svg_phase5_refinement [EXTRACTED]

Communities

Community 0 - “PUMA Master Hub”

Cohesion: 0.07 Nodes (116): AI Use Log, BIB Master APA7, BMAD Agent Roster, PUMA Product Requirements Document (PRD), Carbon Tracking Log, Code Review Log — PUMA, Wilcoxon Signed-Rank Test, Ethics Review Log (+108 more)

Cohesion: 0.07 Nodes (67): Active Reading as Cognitive Practice, BMAD — Business Module Agentic Design, CodeCarbon (CO2 Tracking), CO-STAR Prompt Framework, Chain-of-Thought Prompting (CoT), Design Science Research (DSR), Effort Estimation (Story Points), F1-Macro Evaluation Metric (+59 more)

Community 2 - “Master Bibliography”

Cohesion: 0.04 Nodes (65): Attention Is All You Need (Vaswani et al. 2017), BIB-Master-APA7, CodeCarbon (Courty et al. 2022), Chain-of-Thought Prompting (Wei et al. 2022), Design Science Research (Hevner et al. 2004), How to Read a Paper (Keshav 2007), Kitchenham SLR Guidelines (2007), Llama 3.2 (Meta AI 2024) (+57 more)

Community 3 - “PUMA Project Core”

Cohesion: 0.04 Nodes (64): Human-in-the-Loop (HITL), Reproducibility Failure Taxonomy, Angermeir et al. 2025 — Reproducibility of LLM Studies in SE, LLM Correcting Established Expert Beliefs, Arkani-Hamed et al. 2026 — Gluon Amplitudes (GPT-5.2), APM Autonomy Levels, Assalaarachchi et al. 2026 — Toward Agentic SPM, Software Project Management 3.0 (+56 more)

Community 4 - “SmartPMO Agent System”

Cohesion: 0.05 Nodes (56): GUI Automation Agent, Gao et al. (2023) AssistGUI, AssistGUI Pipeline, Agentic Workflow Authentication, DeBellis et al. (2026) Authenticated Workflows, AIOps Incident Triage, Chen et al. (2024) RCA, LLM Root Cause Analysis (+48 more)

Community 5 - “Tools & Tech Stack”

Cohesion: 0.08 Nodes (52): LoRA and Quantization, KW Category1 AI Agents Swarm Intelligence, KW Category2 Project Management Smart PMO, KW Category3 Research Methodology Academic Rigor, KW Category4 Data Engineering Technical Stack, KW Category5 Evaluation Ethics Sustainability, LN Arora 2024 MASAI, LN Hong 2023 MetaGPT (+44 more)

Community 6 - “Johnny Decimal Index”

Cohesion: 0.06 Nodes (49): Checklists: SLR Quality + Experiment Reproducibility + AI Validation, Glossary Master, Glossary Supplement v2 — Extended Technical Terms, Johnny Decimal Master Index — PUMA Vault, Keywords — Category 1: AI Agents and Swarm Intelligence, Keywords — Category 2: Project Management and Smart PMO, Keywords — Category 3: Research Methodology and Academic Rigor, Keywords — Category 4: Data Engineering and Technical Stack (+41 more)

Community 7 - “Multi-Agent Frameworks”

Cohesion: 0.05 Nodes (47): Agentic Software Engineering, Context Engineering, Spec-Driven Development (SDD), AIDOaRt Framework, AIOpsLab, Mixture-of-Experts (MoE), Switch Transformers, AI Causal Reasoning Limits (+39 more)

Community 8 - “Dev & Infrastructure Tools”

Cohesion: 0.07 Nodes (36): LangGraph Cyclical Workflows, Browser OS — AI Web Browsing Agent, Ollama — Local LLM Inference Engine, Tools — AI Assistants & LLM Models, LN Tools Dev Stack, Tools — Documentation, Organisation & Productivity, Research Tools — Academic Discovery & Management, Web Resources — Blogs, Docs & Key Links (Blog/Web) (+28 more)

Community 9 - “HITL & Ethics”

Cohesion: 0.09 Nodes (32): AI Code Quality Issues, Atomic Human, BDI Agent Architecture, Bounded Autonomy for AI Agents, CO-STAR Framework, Human-in-the-Loop (HITL), LLM Wiki Pattern, Model Context Protocol (MCP) (+24 more)

Community 10 - “Templates & Meta”

Cohesion: 0.11 Nodes (30): Daily Note Template, Dashboard — Experiment Status, Dashboard — Research Progress, Dataview Plugin, Stage 1 Triage — Individual Experiment Notes README, Stage 2 Estimation — Individual Experiment Notes README, Git & GitHub Setup Guide, Log Template (+22 more)

Community 11 - “RAG & Retrieval”

Cohesion: 0.1 Nodes (26): Agentic RAG, Context Engineering, Knowledge Graph, Local LLM (Ollama, LM Studio), Multi-Agent Reinforcement Learning (MARL), RAG — Retrieval-Augmented Generation, Scheduling (MAS), LLM Tokenization (+18 more)

Community 12 - “PM Theory & Motivation”

Cohesion: 0.16 Nodes (23): LN Videos PM Agile AI, MOC Context Engineering, MOC MCP Protocols, MOC PM Agile AI, 12-Factor Agents: Patterns of reliable LLM applications, 3 Advanced AI agent design patterns, Flujos de trabajo utilizando Agentes - Andrew Ng explica, Building AI Agents that actually work (Full Course) (+15 more)

Community 13 - “Research Discovery Tools”

Cohesion: 0.11 Nodes (23): Keshav’s Three-Pass Method — Paper Reading Protocol for PUMA, Design Science Research — Artifact Design Prompts for PUMA, Grounded Theory — Qualitative Analysis of Agent Failures, Semantic Scholar — Core PUMA Topic Search, Semantic Scholar — Task-Specific PUMA Search, Google Scholar — Broad Sweep Queries for PUMA, Consensus — Evidence-Based Search for PUMA, Undermind — High-Precision PUMA Research Query (+15 more)

Community 14 - “AI Knowledge Generation”

Cohesion: 0.11 Nodes (22): BIB Supplement PEC2 AI Knowledge, LN Lu 2024 AI Scientist, LN ArkaniHamed 2026 GluonAmplitudes, LN Degrave 2022 PlasmaControl, LN Felin 2024 Theory Is All You Need, LN Felin 2024 TheoryIsAllYouNeed, LN Jain 2023 GNoME Materials, LN Jain 2023 GNoME Materials (+14 more)

Cohesion: 0.12 Nodes (19): 70 - Archive/71 Old-Notes, MOC Archive, MOC Datasets, MOC Fleeting Notes, MOC Inbox, MOC Library, MOC Meta, MOC Navigation — PUMA Vault (+11 more)

Community 16 - “SmartPMO Vision & Design”

Cohesion: 0.27 Nodes (13): DIRTFT Philosophy, Entrepreneurial Operating System (EOS), Four Decisions Framework, Franchise Prototype, SmartPMO Vision, Systems Mindset, Theory of Constraints, LN-Carpenter-2025-WorkTheSystem (+5 more)

Community 17 - “SDD & BMAD Videos”

Cohesion: 1.0 Nodes (12): LN Videos SDD Specs BMAD, The Official BMad-Method Masterclass (The Complete IDE Workflow), BMAD Method V6: PRD, Architecture, Agents — This Finally Ends Vibe Coding, AgentOS VS BMAD VS Spec Kit: Which AI Development Framework is Worth Your Time?, Plan, Specify, and Implement with Spec Kit, OpenSpec is Insane: Spec Driven Development is 100X Better Than Vibe Coding, BMAD Method Agent Cheatsheet (From Mary to Quinn), Gemini CLI Conductor: Spec-Driven Development 100x Better Than BMAD and OpenSpec (+4 more)

Community 18 - “LLM Agents Videos”

Cohesion: 0.38 Nodes (10): LN Videos LLM Agents MultiAgent, MOC — LLM Agents MultiAgent, Karpathy Just Replaced RAG With Obsidian + Claude Code, CrewAI vs AutoGen vs LangGraph: Features, Pros & Cons Compared (2026), Karpathy’s Autoresearch: We Achieved Near-Human Scores in 2 Hours!, ¡No construyas Agentes IA hasta que veas esto! [Secretos de Anthropic], Orquestación de Agentes: Control Determinista con Hooks, Agentes de IA y LangGraph: cómo las empresas reducen costos (+2 more)

Community 19 - “Observability & Testing Videos”

Cohesion: 0.57 Nodes (8): LN Videos Observability Testing, MOC — Observability Testing, PUMA Evaluation Strategy, Construyendo IA Fiable: Evals, Trazabilidad y Observabilidad — LambdaCast 34, Curso evaluacion LLM con Promptfoo — episodio 1, AI Testing Series Day 2 — Variable Injection & Assertions in promptfoo, Introduction to Observability and Prometheus Tutorial, E124 — Creando agentes con PydanticAI

Community 20 - “Key Book References”

Cohesion: 0.43 Nodes (7): How to Take Smart Notes — Sönke Ahrens (2022), Getting Things Done — David Allen (2001), How Big Things Get Done — Flyvbjerg & Gardner (2023), LN Books Key References, MOC Books, MOC Books Folder, Tasks of MOC Books

Community 21 - “Local LLM & Ollama Videos”

Cohesion: 0.62 Nodes (7): LN Videos Local LLM Ollama, MOC — Local LLM Ollama, PUMA Local Model Deployment, Mac Mini M4 básico para LLM — Probamos modelos de lenguaje para ver sus límites, Ollama vs LM Studio vs Docker Models vs LlamaBarn ¿Cuál es el mejor inferidor?, Run a Powerful AI Agent Locally — Hermes Agent + Ollama (No API Key Required), Claude Code con Modelos Locales: Privacidad y Coste Cero con Ollama

Community 22 - “General AI Tools Videos”

Cohesion: 1.0 Nodes (7): Videos — General AI Tools, SaaS, Automation & Miscellaneous, MOC Miscellaneous Videos, A Visual Tour of Modern LLM Architectures, Supabase con IA (CLI + MCP + Skills): Automatiza todo tu backend, Gemini CLI — Google’s Free Open-Source Coding Agent, CrewAI vs AutoGen vs LangGraph — Features, Pros & Cons Compared (2026), How to Permanently Fix Your Forgetful AI Agents (Full Guide)

Community 23 - “Context Engineering Videos”

Cohesion: 1.0 Nodes (4): LN Videos Context Engineering, Context Engineering Clearly Explained, Ingeniería de Contexto: La Habilidad CLAVE para crear AGENTES de IA ahora mismo, How to build, evaluate, and refine prompts with AI — Latitude

Community 24 - “Swarm Optimization”

Cohesion: 0.5 Nodes (4): Chen et al. (2025) HiveMind, Swarm-Level Agent Optimization, Huang et al. (2024) Internet of Agents, Heterogeneous Agent Protocol

Community 25 - “LangGraph Orchestration”

Cohesion: 0.67 Nodes (3): Arize Phoenix, CrewAI, LangGraph

Community 26 - “Obsidian CSS & Styling”

Cohesion: 1.0 Nodes (2): CSS Snippets & Obsidian Styling Guide, MOC Snippets

Community 27 - “Effort Estimation MOC”

Cohesion: 1.0 Nodes (2): MOC Effort Estimation, README Effort Estimation Papers

Community 28 - “Reproducibility & SE Research”

Cohesion: 1.0 Nodes (2): README Reproducibility SE, MOC Reproducibility SE

Community 29 - “Claude Agent Config”

Cohesion: 1.0 Nodes (1): Claude Agent Configuration

Community 30 - “MOC Concept”

Cohesion: 1.0 Nodes (1): Map of Content (MOC)

Community 31 - “Private Note Template”

Cohesion: 1.0 Nodes (1): Private Note Template

Community 32 - “Idea Template”

Cohesion: 1.0 Nodes (1): Idea Template

Community 33 - “People & Thinkers MOC”

Cohesion: 1.0 Nodes (1): MOC People and Thinkers

Community 34 - “LLM Agents General README”

Cohesion: 1.0 Nodes (1): README LLM Agents General

Community 35 - “AI Knowledge Bibliography”

Cohesion: 1.0 Nodes (1): Bibliography Supplement v3 — Verified New References

Community 36 - “Permanent Note Template”

Cohesion: 1.0 Nodes (1): Template: Permanent Note (concept placeholder)

Community 37 - “Fleeting Note Stub”

Cohesion: 1.0 Nodes (1): PN-concept (Fleeting Permanent Note stub)

Community 38 - “Literature Note Template”

Cohesion: 1.0 Nodes (1): Template: Literature Note (related placeholder)

Community 39 - “Source Note Template”

Cohesion: 1.0 Nodes (1): Template-Source-Note

Community 40 - “NotebookLM Tool”

Cohesion: 1.0 Nodes (1): NotebookLM

Community 41 - “OpenHands Tool”

Cohesion: 1.0 Nodes (1): OpenHands

Community 42 - “Perplexity AI Tool”

Cohesion: 1.0 Nodes (1): Perplexity AI

Knowledge Gaps

190 isolated node(s): Claude Agent Configuration, Daily Review Template, Concepts — Zettelkasten Guide (README), PUMA — LLM Agent Benchmark for ICT Project Management, Johnny Decimal Numbering System (+185 more) These have ≤1 connection - possible missing edges or undocumented components.
Thin community Obsidian CSS & Styling (2 nodes): CSS Snippets & Obsidian Styling Guide, MOC Snippets Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community Effort Estimation MOC (2 nodes): MOC Effort Estimation, README Effort Estimation Papers Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community Reproducibility & SE Research (2 nodes): README Reproducibility SE, MOC Reproducibility SE Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community Claude Agent Config (1 nodes): Claude Agent Configuration Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community MOC Concept (1 nodes): Map of Content (MOC) Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community Private Note Template (1 nodes): Private Note Template Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community Idea Template (1 nodes): Idea Template Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community People & Thinkers MOC (1 nodes): MOC People and Thinkers Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community LLM Agents General README (1 nodes): README LLM Agents General Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community AI Knowledge Bibliography (1 nodes): Bibliography Supplement v3 — Verified New References Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community Permanent Note Template (1 nodes): Template: Permanent Note (concept placeholder) Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community Fleeting Note Stub (1 nodes): PN-concept (Fleeting Permanent Note stub) Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community Literature Note Template (1 nodes): Template: Literature Note (related placeholder) Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community Source Note Template (1 nodes): Template-Source-Note Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community NotebookLM Tool (1 nodes): NotebookLM Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community OpenHands Tool (1 nodes): OpenHands Too small to be a meaningful cluster - may be noise or needs more connections extracted.
Thin community Perplexity AI Tool (1 nodes): Perplexity AI Too small to be a meaningful cluster - may be noise or needs more connections extracted.

PUMA Vault

Explorador

GRAPH_REPORT