This matrix is the repository's completeness contract. It maps each material RAG subsystem to the detailed chapter, executable notebook, reference code, and validation method that covers it. “Executable” means a real, dependency-free algorithm or control contract runs locally. “Analytical” means the handbook documents the method, evidence, equations, trade-offs, and evaluation design but does not pretend to reproduce a large neural model, proprietary service, GPU kernel, or billion-scale index.
No finite review can promise that no paper, product, or future technique exists outside it. The stronger and testable claim is that every lifecycle stage and major technique family in the field map has an explicit home, and every intentionally non-reproduced component is labeled.
1. Scope, history, and research method
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| Definition and boundaries of RAG | Field map §1–2, mathematical primer | 01, complete handbook | models.py, pipeline contracts | Analytical + executable frame |
| Source/task/unit/index/policy taxonomies | Field map §3–13 | 00 | Matrix validation in validate_research.py | Analytical |
| Historical development before the term RAG | Chronology, chronological index | 00, 01 | 203-entry source registry | Primary-source synthesis |
| 2024–2026 frontier and publication-status discipline | Frontier | 00, 07 | sources.json status/date validation | Primary-source synthesis |
| Evidence, claim, chronology, and reproducibility policy | Research method | 00, 08 | Link/source/schema validation | Verified artifact policy |
| Terminology and ambiguous names | Glossary | 00 | Glossary-term threshold | Analytical |
2. Corpus acquisition, parsing, and governance
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| Connectors, crawl/API/file/event ingestion | Corpus §2 | 04 §1–3 | CorpusManifest input contract | Analytical contract |
| Snapshots, CDC, logical IDs, content IDs, versions | Corpus §3, production §1 | 04 §1–3 | CorpusManifest, ManifestEntry, snapshot_hash | Executable |
| Event time, valid time, system/transaction time | Agents/time §21 | 07 §5–6 | TemporalFact, BitemporalStore | Executable |
| HTML/DOM parsing and boilerplate/active content | Corpus §4.1, security §5 | 04 §2, 07 §7 | strip_active_html, injection inspection tests | Partial executable; production parser analytical |
| PDF reading order, layout, OCR, tables, equations | Corpus §4.2–4.4 | 04 §2, 06 §5–7 | Parser evaluation contract | Analytical; no misleading toy PDF parser |
| Slides, spreadsheets, code, image, audio, and video parsing | Corpus §4.5–4.7, structured §13–15 | 06 §7–8 | Modality evidence-unit/citation contract | Analytical |
| Unicode/whitespace normalization | Corpus §5 | 04 §2 | canonicalize_text; exact-offset tests | Executable |
| Exact and near-duplicate policy | Corpus §5.3, security §4 | 04 §2, 07 §9 | ExactDeduplicator, NearDuplicateIndex, near_duplicate_clusters | Executable |
| Metadata schema and immutable provenance | Corpus §6, security §16 | 04 §1, 07 §8–9 | content hashes, HMAC provenance, evidence envelope | Executable teaching contract |
| ACLs, tenant isolation, trust domains | Corpus §14, security §7 | 04 §3, §6, 07 §8 | ACLPolicy, pre-top-k filters, authorize_results | Executable |
| Licensing, copyright, purpose limitation, retention | Security §19–21 | 07 §10 | Governance/release checklist | Analytical/policy-specific |
| Correction, deletion, tombstones, unlearning propagation | Corpus §15, security §20 | 04 §3, 07 §4 | manifest and memory tombstone tests | Executable lifecycle core |
3. Retrieval units and chunking
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| Fixed token windows and overlap | Corpus §7.1 | 01, 04 §4 | text.chunk_document offset/overlap tests | Executable |
| Sentence-aware packing | Corpus §7.2 | 04 §4 | sentence_chunks exact character/token lineage | Executable |
| Paragraph/section/DOM-structure chunks | Corpus §7.3 | 04 §4 | section_chunks, Markdown path tests | Executable for Markdown; other parsers analytical |
| Semantic boundary and discourse chunking | Corpus §7.4 | 04 §4 | Evaluation protocol and substitution seam | Analytical/model-dependent |
| Proposition/atomic-fact indexing | Corpus §7.5 | 04 §4 | Claim/evidence selection interfaces | Analytical/model-dependent extraction |
| Parent–child/small-to-big retrieval | Corpus §7.6, structured §10 | 04 §5, 06 §4 | parent_child_chunks, hierarchy selection tests | Executable |
| Contextual prefixes, late chunking | Corpus §7.7, retrieval §5–6 | 04 §4, §7 | Embedding contract and ablation design | Analytical/model-dependent |
| Recursive/hierarchical summaries | Corpus §7.8, structured §10 | 06 §4 | HierarchyNode, leaf-coverage selector | Executable selector; learned summaries analytical |
| Table rows/cells, page regions, code symbols, media segments | Corpus §7.9, structured §11–15 | 06 §5–8 | TableRow; modality provenance matrix | Table retrieval executable; others analytical |
| Chunk boundary/recall/duplication/cost evaluation | Corpus §8 | 04 §4–5, 08 §5 | lineage, token budgets, evidence flow | Executable |
4. Indexes and first-stage retrieval
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| Boolean, phrase, proximity, analyzers | Retrieval §3.1 | 04 §6 | Inverted-index execution substrate | Analytical beyond unigram lab |
| TF–IDF/vector-space and query likelihood | Chronology, retrieval §3.2, math §13 | 01 | formula/tests via retrieval metrics | Analytical + adjacent executable |
| BM25, BM25F, BM25 variants | Retrieval §3.3–3.4, math §2 | 01, 04 §6 | scan BM25 and postings-list InvertedIndex | Executable BM25; fielded variants analytical |
| Pseudo-relevance feedback/Rocchio/relevance models | Retrieval §3.5, math §13 | 05 §5 | Transformation experiment contract | Analytical |
| WAND/Block-Max WAND and top-k pruning | Retrieval §3.6 | 04 §6 | Postings exposed; exact semantics documented | Analytical optimization |
| DeepCT/doc2query/DeepImpact/uniCOIL/COIL | Retrieval §4.1–4.3 | 05 §4 | Source/evaluation coverage | Analytical neural models |
| SPLADE and learned sparse regularization | Retrieval §4.4, training §7 | 04 §6, 05 §4 | Loss/index-growth discussion and metrics | Analytical neural model |
| Dense dual encoders: DPR, ANCE, Contriever, GTR, E5, DRAGON | Retrieval §5, chronology | 01, 04 §7, 05 §4 | exact cosine index; objective calculations | Executable interface/math; neural checkpoints analytical |
| Late interaction: ColBERT, PLAID, XTR, CITADEL | Retrieval §6, math §4 | 06 §6 | late_interaction_score, pooling tests | Executable MaxSim core |
| Exact dense search oracle | Corpus §12 | 04 §7–8 | ExactCosineIndex | Executable |
| IVF coarse quantization | Corpus §13.2, math §17 | 04 §8 | IVFCoarseIndex, exact recall audit | Executable |
| PQ/OPQ/scalar quantization | Corpus §13.3, math §17 | 04 §9 | ScalarQuantizer, ProductQuantizer, distortion/storage audit | Executable PQ/scalar; OPQ analytical |
| HNSW | Corpus §13.4, math §18 | 04 §8 | Algorithm/evaluation contract | Analytical; no toy scale claim |
| DiskANN/Vamana, SPANN, ScaNN | Corpus §13.5–13.7 | 04 §8 | Resource/recall evaluation design | Analytical systems algorithms |
| ANN recall, filters, latency, storage, updates | Corpus §13–14, retrieval §7 | 04 §8–10 | evaluate_ann_recall, pre-top-k filters, quantization audit | Executable |
5. Query understanding, fusion, and ranking
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| Intent classification and entity/temporal/metadata extraction | Retrieval §2 | 05 §5, 07 §2 | AdaptiveRetriever routing | Executable proxy + analytical production design |
| Spelling/normalization/aliases | Retrieval §8.1 | 01, 05 §5 | canonical/synonym transforms | Executable basic transform |
| Multi-query/RAG-Fusion | Retrieval §8.3 | 02, 05 §5–6 | MultiQueryRetriever, RRF | Executable |
| HyDE, Query2Doc, generated expansion | Retrieval §8.3–8.4 | 05 §5 | Ablation/logging contract | Analytical model-dependent generation |
| Conversation rewriting, decomposition, step-back | Retrieval §8.5–8.7 | 07 §1–2 | iterative planner/trace | Executable simple decomposition; learned rewriting analytical |
| RRF | Retrieval §9.1, math §7 | 01, 05 §6 | reciprocal_rank_fusion, HybridRetriever | Executable |
| Score calibration, CombSUM/CombMNZ | Retrieval §9.2–9.4, math §16 | 05 §6 | min-max/z-score calibrated CombSUM | Executable CombSUM; learned/calibrated variants analytical |
| Cross-encoder, monoT5/RankT5, RankGPT/listwise reranking | Retrieval §10, training §9 | 02, 05 §7 | CrossFeatureReranker; point/pair/listwise objectives | Executable boundary/math; neural rerankers analytical |
| MMR/deduplication/diversity | Retrieval §11.1, context §2–3 | 02, 05 §8 | ContextPacker | Executable |
| Weighted set cover/knapsack/claim coverage | Retrieval §11.2, math §19 | 05 §8 | greedy_budgeted_coverage, selection regret | Executable |
| Authority, independence, time, contradiction-aware selection | Retrieval §11.3–11.5 | 05 §8, 07 §5–9 | trust domains, bitemporal lookup, duplicate clusters | Executable primitives + analytical policies |
| Sufficiency/calibration/stopping | Retrieval §12, context §17 | 02, 07 §2 | abstention thresholds, bounded stop conditions | Executable proxy + analytical calibration |
6. Context and generation
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| Context budgeting, deduplication, ordering, serialization | Context §1–5 | 02, 05 §8–10 | ContextPacker, untrusted evidence envelope | Executable |
| Extractive, abstractive, token, and latent compression | Context §6 | 06 §4, §6 | hierarchy/vector pooling; evaluation contract | Partial executable; learned compressors analytical |
| Prompt-based context-only/permissive RAG | Context §7 | 02 | extractive cited generator | Executable policy proxy |
| Original RAG-Sequence/RAG-Token latent integration | Context §8, chronology, math §5 | 01 | formal objective and historical result audit | Analytical neural architecture |
| FiD/multi-passage readers | Context §9, chronology | 01 | mechanism/evaluation coverage | Analytical neural architecture |
| REALM, RETRO, Atlas, kNN-LM/retrieval pretraining | Context §10, chronology | 00, 01 | objectives, dates, results, limits | Analytical neural systems |
| Frozen/black-box generator RAG | Context §11 | 02 | modular pipeline seam | Executable interface |
| Iterative generation: IRCoT, FLARE, ITER-RETGEN, Self-RAG | Context §12, agents §5–11 | 07 §1–2 | budgeted iterative trajectory | Executable control proxy + analytical methods |
| Long-form claim decomposition and citations | Context §14–15, evaluation §6 | 02, 08 §2–5 | source-linked Citation, claim/citation metrics | Executable basic attribution + analytical NLI/human audit |
| Verification/correction loops | Context §16 | 02, 08 §5 | trace/failure attribution | Partial executable |
| Abstention, uncertainty, conflicts, numerical and temporal claims | Context §17–18 | 02, 07 §5–6 | pipeline abstention and temporal ambiguity | Executable core |
| Structured/constrained outputs | Context §19, production §10 | 08 §1 | schema/version/retry contract | Analytical |
| Long context versus retrieval and Self-Route | Context §21, field map §14 | 07 §2 | route evaluation contract | Analytical |
7. Learning and optimization
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| Positive-label provenance and bias | Training §1–3 | 05 §1 | label contract/checklist | Analytical |
| Random/in-batch/cross-batch/BM25/ANN/teacher/adversarial negatives | Training §4 | 05 §2–3 | NegativeExample, mine_hard_negatives | Executable policy core |
| False-negative masking | Training §4.8 | 05 §2 | false_negative_mask, masked loss tests | Executable |
| InfoNCE/multiple-negative softmax | Training §5.1, math §14 | 05 §1–2 | contrastive_loss, in_batch_contrastive_loss | Executable math |
| Triplet/hinge/pairwise logistic/listwise losses | Training §5.2–5.4 | 05 §1, §4 | hinge, logistic, listwise implementations | Executable math |
| Margin-MSE and KL/listwise distillation | Training §5.5–5.6, math §15 | 05 §4 | kl_distillation_loss | Executable KL core; margin-MSE analytical |
| Retrieval-oriented pretraining and learned sparse/multi-vector training | Training §6–8 | 05 §4 | primary-source/objective audit | Analytical neural training |
| Reranker/rewriter/decomposer training | Training §9–10 | 05 §5–7 | ranking objectives and trace evaluation | Executable objectives; models analytical |
| Generator SFT, evidence dropout, citation training | Training §11 | 05 §9–10 | curriculum/release protocol | Analytical neural training |
| Latent-document joint learning, EM, reader-to-retriever distillation | Training §12–14, math §5 | 00 | objective derivations | Analytical neural training |
| DPO/preference optimization | Training §15 | 05 §9 | dpo_loss | Executable math |
| RL search: policy gradients, value learning, process/outcome rewards | Training §16–17, agents §11–12 | 05 §9, 07 §1–2 | returns/advantages/REINFORCE, bounded agent | Executable math/control; large RL runs analytical |
| Reward hacking, curriculum, domain/continual learning | Training §17–21 | 05 §9–10 | adversarial/release checklist | Analytical |
8. Structured and multimodal RAG
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| Curated KG entity linking/path retrieval/GNN QA | Structured §3 | 06 §1–2 | weighted PPR and graph expansion | Executable propagation; neural/entity linking analytical |
| Passage/entity associative graphs and HippoRAG | Structured §4 | 02, 06 §1–2 | metadata graph expansion, PPR | Executable core |
| Microsoft GraphRAG community reports/DRIFT | Structured §5 | 06 §3 | community-selection proxy and cost/eval contract | Analytical LLM extraction/summarization |
| Query-specific/dynamic graph RAG | Structured §6 | 06 §1–2 | query-seeded graph propagation | Executable core |
| PageRank, path, beam, community, GNN, LLM traversal | Structured §7 | 06 §1–3 | exact weighted PPR; other algorithms documented | Partial executable |
| Graph extraction/entity resolution/confidence/contradiction | Structured §8–9 | 06 §1–3 | construction/evaluation contract | Analytical/model-dependent |
| RAPTOR/structural hierarchy/multi-resolution/map-reduce | Structured §10 | 06 §4 | select_hierarchy with leaf lineage | Executable selector; learned summaries analytical |
| Table serialization, row/cell retrieval, SQL/program execution | Structured §11 | 06 §5 | retrieve_table_rows with numeric/schema features | Executable row retrieval; governed SQL analytical |
| Visual documents: text-first, captions, ColPali, VisRAG, VDocRAG | Structured §12 | 06 §6 | patch MaxSim and pooling/storage trade-off | Executable core; VLMs analytical |
| Image/chart/map/audio/video RAG | Structured §13 | 06 §7 | modality evidence/citation contract | Analytical |
| Code/repository RAG | Structured §14 | 06 §8 | symbol/graph/commit provenance design | Analytical |
| Web/live search and API/tool RAG | Structured §15, agents §13 | 06 §8, 07 | source-time/tool permission contracts | Analytical |
| Multilingual/cross-lingual RAG | Structured §16 | 06 §8 | language-pair evaluation matrix | Analytical |
| Biomedical, legal, finance, science, enterprise/personal | Structured §17–21 | 06 §8 | domain-specific authority/time/privacy gates | Analytical |
9. Adaptive search, agents, memory, and time
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| No/one/multi retrieval routing and Adaptive-RAG | Agents §3–4 | 07 §2 | AdaptiveRetriever | Executable deterministic proxy |
| Self-RAG reflection/control tokens | Agents §5, frontier | 07 §1–2 | control taxonomy and trace evaluation | Analytical neural model |
| Corrective/evidence-grading RAG | Agents §6 | 07 §1–2 | retry/stop policy seam | Partial executable |
| FLARE/uncertainty-triggered active retrieval | Agents §7 | 07 §1–2 | bounded policy contract | Analytical neural uncertainty |
| ReAct, Self-Ask, IRCoT and multi-hop state | Agents §8–10 | 07 §1 | BudgetedIterativeRetriever, plan/accumulate/stop trace | Executable control proxy |
| Search-R1, ReSearch, StepSearch, GRIP, Q-RAG, DeepRAG, HiPRAG | Agents §11–12, frontier | 07 §1–2 | reward/control comparison | Analytical peer-reviewed frontier |
| Source/retriever/tool/long-context routing | Agents §13–14 | 07 §2 | router and regret/route metrics | Executable proxy + analytical policies |
| Working/episodic/semantic/procedural/profile/latent memory | Agents §15–16 | 07 §3 | typed MemoryRecord kinds | Executable external types; latent memory analytical |
| Memory write/retrieve/update/supersede/consolidate/forget | Agents §17–19 | 07 §3–4 | MemoryStore, write policy, groups, tombstones | Executable |
| LongMemEval and memory metrics | Agents §20 | 07 §3–4 | lifecycle metric contract | Analytical benchmark application |
| Temporal scoring/version conflict/freshness | Agents §21–22, math §20 | 07 §5–6 | bitemporal lookup, decay, stale rate | Executable |
| Temporal/permission/model-sensitive caches | Agents §23, production §9 | 07 §6, 08 | cache_identity | Executable key contract |
10. Grounding, evaluation, and statistics
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| Retrieval precision/recall/MRR/nDCG | Evaluation §2, math §11 | 01, 03, 08 §2 | metric implementations/tests | Executable |
| Answer EM/token F1/claim correctness/completeness | Evaluation §3 | 03, 08 §3 | EM/F1; claim evaluation design | Partial executable |
| Faithfulness/context utilization | Evaluation §4 | 03, 08 §2–3 | lexical faithfulness explicitly labeled proxy | Executable proxy + analytical human/NLI audit |
| Citation precision/recall/entailment/completeness/authority | Evaluation §5–6, context §14–15 | 02, 03, 08 | source-linked citations, precision/recall | Partial executable; entailment/authority human audit analytical |
| Abstention/selective prediction/calibration | Evaluation §7, context §17 | 02, 03 | abstention behavior/tests | Executable proxy |
| RAGAS, ARES, RAGChecker, RAGTruth/RAGBench | Evaluation benchmark audit | 08 §2 | taxonomy, assumptions, use guidance | Analytical evaluator audit |
| KILT, BEIR, MTEB/MMTEB, BRIGHT | Evaluation benchmark audit | 08 §2 | benchmark/metric compatibility map | Analytical benchmark audit |
| RGB, CRUD-RAG, CRAG, NoMIRACL, mtRAG, TREC RAG | Evaluation benchmark audit, frontier | 08 §2 | targeted risk-suite guidance | Analytical benchmark audit |
| Oracle/closed-book/distractor/end-to-end controls | Evaluation protocol | 08 §3 | layered experiment design | Executable end-to-end; oracle setup analytical |
| Per-query slices and failure attribution | Evaluation, retrieval §15 | 03, 08 §4–5 | metrics_by_tag, evidence_flow | Executable |
| Paired bootstrap intervals and human/judge audit | Evaluation statistics | 03, 08 §4 | paired_bootstrap_delta | Executable bootstrap; human/PPI analytical |
11. Security, privacy, and governance
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| Assets, actors, trust boundaries, threat model | Security §1–3 | 07 §7–10 | explicit untrusted evidence envelope | Analytical + executable boundary |
| Corpus/memory poisoning and trigger backdoors | Security §4 | 07 §9 | near-duplicate clusters and canaries | Executable diagnostics; attacks/guarantees analytical |
| Indirect prompt injection and tool coercion | Security §5 | 07 §7 | active-content removal, multi-signal inspection | Executable detector; explicitly not a guarantee |
| Retrieval manipulation/source spoofing | Security §6 | 07 §8–9 | trust domains, signed provenance | Executable teaching controls |
| Cross-tenant leakage and ACL bypass | Security §7 | 07 §8 | tenant/ACL denial tests | Executable |
| Corpus extraction/membership inference/embedding-query privacy | Security §8–10 | 07 §10 | threat/evaluation matrix | Analytical privacy research |
| PIR, encrypted/TEE/federated privacy-preserving retrieval | Security §11 | 00 | design/performance trade-off audit | Analytical cryptographic/systems methods |
| Multimodal/graph privacy and cache side channels | Security §12–13 | 07 §10 | threat and cache-identity contracts | Analytical + executable cache key |
| Safety degradation and RAG-specific red teaming | Security §14, §25 | 07 §7–10 | attack matrix/checklist | Analytical |
| Resource/DoS attacks | Security §15, production §7–8 | 07 §7, 08 §7 | oversized-content signal and hard budgets | Executable basic controls |
| Citation/provenance integrity | Security §16 | 07 §8–9 | hashes, HMAC, immutable IDs | Executable teaching controls |
| Filtering, conflict-aware, certified/conformal defenses | Security §17–18, frontier | 07 §9 | assumptions/limits and aggregation diagnostics | Analytical guarantees; executable diagnostics |
| Supply chain, logging privacy, secure release | Security §22–24 | 07 §10, 08 §9–10 | release fingerprint and governance gates | Executable release identity + analytical controls |
| Incident response and residual human control | Security §26–27, production §25 | 07 §10, 08 §10 | runbook checklist | Analytical/organization-specific |
12. Production systems and economics
| Subject | Detailed material | Notebook | Executable or validation evidence | Coverage level |
|---|
| Data/ingest/control/evaluation/serving planes | Production §1 | 08 §1 | versioned component contracts | Analytical architecture |
| Auth → route → retrieve → rerank → pack → generate → verify request path | Production §2 | 02, 08 §1 | pipeline TraceEvents | Executable |
| Quality/latency/availability/freshness/security/cost SLOs | Production §3 | 08 §6–8 | p50/p95/p99, cost/failure/cache summaries | Executable |
| Capacity model and queuing limits | Production §4 | 08 §6–7 | calls/tokens/latency budget model | Partial executable |
| Sparse/dense/multi-vector/federated serving | Production §5 | 04, 06 | local reference indexes; production topology analytical | Partial executable |
| Sharding, replicas, batching, scheduling, backpressure | Production §6–7 | 08 §6–7 | SLI/capacity design | Analytical distributed systems |
| Context/tool/token/time/cost hard budgets | Production §8 | 07 §1–2, 08 §7 | ServiceBudget, check_budget, bounded agent | Executable |
| Source/parse/embedding/retrieval/answer/KV caches | Production §9 | 07 §6, 08 §6 | cache identity/hit metrics | Executable key/metrics; cache engines analytical |
| Generation serving, prefill/decode, speculative retrieval | Production §10 | 08 §6 | stage metrics/evaluation contract | Analytical serving kernels |
| Dependencies, retries, circuit breakers, degraded modes | Production §11–12 | 08 §7, §10 | budget/rollback/runbook contract | Analytical |
| Trace schema and online monitoring | Production §13–15 | 08 §1, §6 | RequestMeasurement, StageMeasurement, stage summaries | Executable |
| Query/corpus/model/index/evaluator drift | Production §16 | 08 §4, §9–10 | slice/release comparison protocol | Analytical + executable primitives |
| Offline/shadow/canary/rollback and feature flags | Production §17–18 | 08 §9–10 | content-addressed release manifest | Executable identity; deployment analytical |
| End-to-end cost accounting | Production §19, math §21 | 08 §6–8 | per-stage/request cost | Executable |
| Efficiency: datastore, pruning, ANN, pooling, batching, early stop | Production §20, frontier | 04 §8–9, 06 §6, 08 §6–8 | recall/storage/latency/cost trade-offs | Executable representative levers |
| Vector/search platform and build-vs-buy selection | Production §21–22 | 08 §8–10 | requirement/Pareto framework | Analytical/vendor-neutral |
| Multi-region/DR/testing/runbooks/readiness | Production §23–29 | 08 §9–10 | release and incident checklists | Analytical/organization-specific |
| Constrained utility and Pareto frontier | Production §19–20, math §21 | 08 §8 | SystemCandidate, pareto_frontier, constrained_choice | Executable |
13. Verification inventory
The current automated gate checks:
- all handbook and notebook artifacts exist and internal links resolve;
- the source registry has unique IDs/URLs, valid statuses, dates, topics, and a minimum breadth threshold;
- the handbook exceeds a minimum narrative/primary-link threshold and contains no unfinished-marker placeholders;
- the complete field-map topics appear in this matrix;
- every required notebook is valid nbformat 4, uses a Python kernel, has been executed, contains no error output, and retains saved output;
- corpus documents preserve dates, resolvable sources, and graph metadata;
- unit/integration tests cover retrieval, chunk lineage, indexes, quantization, training objectives, fusion/selection, graph/table/visual primitives, memory, bitemporal lookup, security controls, SLOs, budgets, citations, and abstention;
- notebook regeneration and execution are deterministic;
- source distributions, research word count, notebook cell counts, glossary breadth, and coverage-row count are printed for inspection.
The intentionally non-reproduced surfaces—large neural training, proprietary models, distributed search engines, cryptographic private retrieval, and billion-scale benchmarks—are documented with their objectives, assumptions, primary evidence, costs, and a replacement/evaluation contract. A small local simulation is never presented as evidence for their published performance.