TT Claude Memory

project_z_research_programme_24_files

Properties1
description2026-04-22 Rich committed to a 24-Z-file library-grounded research programme to stress-test the locked v2 architecture (Ⅲ + D-ii + D-jurisdictions + α_max + 7-module + ε admin) before build commences. Six waves — Foundations, Primary, Practice, Cross-cutting, Consolidation, Adversarial-critique. ~6-7 hours wall-clock across multiple agent-launch messages. Designed for watertight evidence-grounding, not option-generation.

5 min read

Context: Rich stated 2026-04-22 “we now have our most solid plan yet, so it is worth spending time to ensure our plan is watertight.” This shifted the research posture from speed to thoroughness.

The 24 Z-files

Original 10 (pre-tagged Scale-3; Z1-Z5 have existing notes files with ~200 ULID-cited quotes)

  1. Z1 Z-PARADIGM — JSON Schema / JSON-LD / SPARQL / SHACL / OWL 2 DL / Catala stack coherence
  2. Z2 Z-RULEBASE — Catala statutory + interpretive scope + formal methods (Alloy, TLA+, Carneades)
  3. Z3 Z-AGENT — MCP + Cloudflare AI Gateway + Haiku-default + agent-readiness + Scenario-2
  4. Z4 Z-LICENCE — Apache / AGPL / dual-commercial / CLA / content-type-per-row licensing
  5. Z5 Z-COMMERCIAL — subscription (IW), one-off + Living Monitoring (MFI), AI-vendor tier, freemium, acquirer narrative
  6. Z6 Z-UK-LAW — E&W IHTA + AEA 1925 + Wills Act 1837 + I(PFD)A + BPR/APR/GWR/QSR + trust IHT + capacity
  7. Z7 Z-FAITH — Islamic (UK-Sunni-Hanafi-Deobandi + madhhabs) + halakhic + Hindu + cultural-sensitivity + reification-mitigations
  8. Z8 Z-SCOPE — Scotland (jus relictae, legitim, Sheriff Court) + Switzerland (ZGB + cantonal) + Brussels IV + conflict-of-laws
  9. Z9 Z-STANDARDS — ISO/OASIS/FHIR/ELI/ECLI/Akoma Ntoso/W3C Recs + OASIS TC → JTC 1 PAS fast-track
  10. Z10 Z-CAPACITY — mental capacity (MCA 2005 / Banks v Goodfellow) + fiduciary duties + facility-and-circumvention (Scots)

New 8 (forced by v6.6 discovery + asset-catalogue scope)

  1. Z11 Z-ASSETS — Asset × 16 categories × AssetCollection × Space × Valuation × IHTA statutory-value-types × condition grading × per-asset + per-field provenance. Assesses Shape-C Assets partner-model.
  2. Z12 Z-MARKETPLACE — three-role model (collector/dealer/public) + offer state machine + escrow + matching engine + Stripe Connect + wishlist. Shape-B Catalogue category-sponsorship marketplace.
  3. Z13 Z-PROBATE — Estate state-machine + executor duties + non-probate transfers + lifetime transfers
  4. Z14 Z-TRUSTS — express/implied/constructive/resulting + Trustee Act 2000 + power-of-appointment + discretionary IHT + Pitt v Holt + proxy-authorisation
  5. Z15 Z-DELEGATION — W3C VC 2.0 + OIDF DASDT + OAuth 2.0 Token Exchange + Kantara UMA + posthumous-AI + RUFADAA + digital-estate-service
  6. Z16 Z-INTEGRATIONS — Clio + Actionstep + Wealth.com + Addepar + Settld + Estateably + LegalZoom + LEAP + PracticePanther + Rocket Lawyer
  7. Z17 Z-ONTOLOGY — OWL 2 DL legal-domain engineering — FIBO + LKIF-Core + Baader/Horrocks + Hitzler/Krötzsch/Rudolph + Pellet/HermiT
  8. Z18 Z-GOVERNANCE — Founding Steward + consortium handover vehicle (OASIS TC / TT-founded 501c3 / W3C CG) + FHIR three-tier precedent + partner-tier badging

Optional 5 accepted (watertight-plan deepening)

  1. Z19 Z-HEALTH — FHIR interop + death registration (DWP/HMRC) + bereavement notification + Tell-Us-Once + NHS records
  2. Z20 Z-FINANCE — open banking + pensions (PPF + trust-based) + investment-platform mapping + wealth-manager APIs + HMRC direct-integration
  3. Z21 Z-IDENTITY — Person entity + DID + organisation identifiers (LEI, SRA, Wikidata, GLEIF) + practitioner credential verification
  4. Z22 Z-TEMPORAL — bi/tri/tetra-temporal modelling + Parry-Kerridge 4-D time-lattice + Kleppmann bi-temporal + 80-year replay
  5. Z23 Z-CONFORMANCE — three-level conformance testing (Schema Valid / Referentially Intact / Jurisdiction Complete) + certification programme + v6.6’s 2,100+ test-case inheritance

New 24th (added 2026-04-22 per Rich’s “module decomposition drift” observation)

  1. Z24 Z-MODULARITY — INHERIT-Core + 6-families split resolution. Answers A4.38 (Core as explicit 7th module — LOCKED as A), A4.39 (per-family commercial packaging), X9.15 (Core separation confirmed), X9.16 (InheritKit repo structure). Library evidence against FHIR-Foundation + FIBO-EDMC + iText-playbook + Meeker open-core + per-partner-per-family three-shape model.

Six-wave dispatch structure

WaveZ-filesPurposeWall-clock
1 FoundationsZ1-PARADIGM, Z6-UK-LAW, Z8-SCOPE, Z9-STANDARDS, Z11-ASSETS, Z18-GOVERNANCESubstrate findings; dependency-free~60-90 min (6 parallel)
2 Primary layersZ2-RULEBASE, Z7-FAITH, Z10-CAPACITY, Z13-PROBATE, Z14-TRUSTS, Z17-ONTOLOGYRule-layer + ontology-layer (depend on W1)~60-90 min (6 parallel)
3 Practice + operationsZ3-AGENT, Z4-LICENCE, Z5-COMMERCIAL, Z12-MARKETPLACE, Z15-DELEGATION, Z16-INTEGRATIONSCommercial + operational (depend on W1+W2)~60-90 min (6 parallel)
4 Cross-cuttingZ19-HEALTH, Z20-FINANCE, Z21-IDENTITY, Z22-TEMPORAL, Z23-CONFORMANCE, Z24-MODULARITYOrthogonal concerns (depend on W1-W3)~60-90 min (6 parallel)
5 ConsolidationAll 24 — cross-reference, contradiction register, gap registerIntegration~45-75 min (1 agent)
6 Adversarial critiqueAll 24 + consolidation — break-the-plan agentRed-team stress-test~45-75 min (1 agent)

Total: ~26 agent-runs across 6-7 hours wall-clock. Agents run dependency-ordered (later waves can cite earlier findings).

Why staged, not single-burst

  • Dependencies exist — Z-ONTOLOGY depends on Z-PARADIGM; Z-MARKETPLACE depends on Z-ASSETS; Z-FAITH depends on Z-UK-LAW + Z-SCOPE + Z-CAPACITY
  • Gaps only appear in cross-reference — single-burst agents can’t see each other’s work
  • Watertight requires adversarial review — Wave 6 is where real stress-test happens
  • Wave-3+ agents can cite Wave-1+2 findings, enriching evidence graph

Z1-Z5 disposition (DEFERRED)

Three options still on table (Rich to decide):

  • (a) Re-dispatch all 24 including refresh of Z1-Z5 under expanded scope
  • (b) Dispatch only 19 new (Z6-Z24); existing Z1-Z5 notes stay as-is
  • (c) Hybrid — dispatch 19 new + one staleness-review agent that flags which Z1-Z5 findings need refresh under Ⅲ + D-ii + α_max

My recommendation: (c) hybrid — cheapest, evidence-driven, doesn’t duplicate ~200 ULID-cited quotes still valid.

Preparatory commits needed before Wave 1 launches

  1. architecture-proposal v1.5 committed (Ⅲ + D + α_max + mitigations + 7-module + ε+δ + 30-48 mo)
  2. 24 Z-stub notes files created at ~/off-github/library/projects/inherit/Z{N}-notes.md
  3. 24 research-request YAML blocks added to ~/off-github/library/research-requests.md
  4. Memory updates: this file + supersede project_scotland_as_second_jurisdiction.md + new project_v2_architecture_locked_2026_04_22.md + new project_partner_model_three_shapes.md

Each research agent produces

Notes file at ~/off-github/library/projects/inherit/Z{N}-notes.md with:

  • Requirements this spine addresses (from v1.5 + relevant mega-spec claims)
  • Library evidence organised by sub-topic (ULID-cited quotes with backlinks)
  • Findings contradicting current plan (adversarial)
  • Gaps + open questions
  • Cascade recommendations (specific claim edits)
  • Cross-references to other Z-files

Session-continuation rule

Waves can run across multiple sessions — each wave commits its artefacts before the next launches. No single-session requirement.

Graph View

Backlinks