project_technology_research_discoveries_catalogue_2026_04_25

Technology research discoveries catalogue 2026-04-25

Created at Rich’s request: “please save details of all the discoveries we have made.” Comprehensive audit trail of the dramatic research hour that followed from the Catala ecosystem-discovery precedent (11th session auto-deep-dive win).

Programme architecture

13 T-files researched (complete):

• T1 MCP / T2 W3C VC 2.0 + DID + KERI / T3 JSON Schema + Sourcemeta / T4 SHACL — HIGH-velocity batch (Rich-directed deep-dives)
• T5 SKOS+XKOS+cmns-cls / T6 PROV-O / T7 Akoma Ntoso+ELI+ECLI / T8 ROBOT — Tier-1 MEDIUM batch
• T9 OpenAPI+Arazzo / T10 SSSOM / T11 F*+Mopsa / T12 External ontologies (CIDOC/FIBO/LADM/BFO) — Tier-2 batch
• T13 LinkML — scope-shifting Phase-1.5 spike candidate

6 T-files Round 2 dispatched 2026-04-25 (pending):

• T14 OSCAL / T15 xpSHACL-ecosystem / T16 Jentic JAIRF / T17 OntoGPT / T18 Biolink Model / T19 Gaia-X

Total programme coverage when Round 2 returns: 19 of 19 technologies researched.

---

T1 MCP — 10 discoveries (HIGH InheritKit / MEDIUM INHERIT v2)

1. MCP donated to Linux Foundation (AAIF) Dec 9, 2025 — co-founded by Anthropic + OpenAI + Block; support from Google + Microsoft + AWS + Cloudflare + Bloomberg. Removes Anthropic lock-in risk.
2. Spec 2025-11-25 stable; SDK v2 Q1 2026 anticipated — no breaking v2 yet.
3. 97M monthly SDK downloads, 10k active servers, first-class client support across ChatGPT / Claude / Cursor / Gemini / MS Copilot / VS Code.
4. Streamable HTTP replaced HTTP+SSE (March 2025) — SSE-only servers deprecating mid-2026.
5. Tasks primitive (SEP-1686) — call-now/fetch-later durable state machine. THE primitive for catala-explain .docx delivery.
6. Server Cards via .well-known/mcp/server-card.json + official MCP Registry live (Anthropic + GitHub + Microsoft).
7. Cloudflare MCP Gateway production-ready April 2026 — compounds Scenario-2 lock.
8. Arazzo + MCP = industry-consensus pattern — workflows AS MCP tools, not raw endpoints.
9. OAuth 2.1 + PKCE mandatory; DPoP + WIF SEPs sponsored — VC-over-DPoP enables VC-in-MCP-auth.
10. No succession/probate MCP exists — InheritKit enters open category.

T2 W3C VC 2.0 + DID 1.1 + KERI — 13 discoveries (HIGH)

1. VC 2.0 family = W3C Rec 15 May 2025 — 7 specs ratified together (Data Model + VC-JOSE-COSE + Data Integrity + EdDSA/ECDSA + BitstringStatusList + Controlled Identifiers).
2. Two securing pathways — Data Integrity (JSON-LD) vs SD-JWT VC (JOSE) — both needed.
3. Bitstring Status List v1.0 — revocation infrastructure production-ready.
4. JSON-LD Context stability is load-bearing — 7 A-43 VC schemas need TT-hosted stable URIs.
5. DID 1.1 CR 5 March 2026 — layered on CID v1.0.
6. did:web + did:peer + did:key recommended; ledger methods deprioritised.
7. KERI CROSSED INTO PRODUCTION — GLEIF vLEI live + Trust-over-IP ratified KERI + ACDC + CESR Jan 21, 2026. Upgrades KERI from Phase-2+ to Phase-1-eligible.
8. ACDC = VC variant with native credential-chaining DAG via e field + SAID content-addressing — perfect for delegation chains.
9. No production precedent for Catala-backed VCs — INHERIT first-mover W3C VC CG contribution.
10. MCP auth = OAuth 2.1 pure — VC→OAuth bridge at IdP via RFC 8693.
11. EUDI ARF 2.7.3 mandates SD-JWT VC + ISO 18013-5 mdoc for Switzerland — Phase-1 constraint (not Phase-2 deferrable).
12. UK DIATF gamma 0.4 statutory 1 Dec 2025 — W3C VC + OpenID4VC + JWT accepted.
13. Four-layer credential stack framework (Identifier/CID + VC + Securing + Status) + two protocol layers (OpenID4VCI/VP + VC→OAuth bridge).

T3 JSON Schema + Sourcemeta — 8 discoveries + 🚨 urgent commercial finding (HIGH)

1. JSON Schema abandoned IETF; v1/2026 is new stable-evolving spec — pin Phase-1 to v1/2026 not 2020-12.
2. JSON-LD WG charter runs to Jan 2028; NO 1.2 in charter — pin 1.1.
3. Blaze validator peer-reviewed (arXiv 2503.02770, 2025) — 10× faster + 100% Bowtie compliant.
4. Sourcemeta jsonschema CLI mature npm/pip-style schema package manager.
5. OpenAPI 3.2 (Sep 2025) ratifies 3.1 unification + adds jsonSchemaDialect — INHERIT can publish own dialect URI.
6. Catala natively supports JSON Schema for scope input/output — law-vs-practice pre-wired.
7. LinkML most mature single-source authoring → T13 spike candidate.
8. Viotti’s own admission peer-reviewed: “JSON Schema is not a modeling language, but a constraints language” (Itelman Ch 8 p.130).

🚨 URGENT: Sourcemeta One MCP support “coming soon” — overlaps with TT’s InheritKit revenue thesis. Direct Juan Cruz Viotti conversation needed BEFORE Sourcemeta ships MCP.

T4 SHACL — 9 discoveries (HIGH)

1. SHACL 1.2 is multi-deliverable family (6 specs); W3C WG charter to end-2026; Core 1.2 Rec unlikely before late 2027.
2. pySHACL v0.31 (Jan 2026) + KonradHoeffner/shacl GitHub Action = Phase-1 CI tool.
3. TopBraid SHACL API v1.5.0 (March 2026) — European Commission validators.
4. Jena ≈ Stardog > RDF4J (2.4× slower); GraphDB 11.x has rare incremental-validation.
5. Re-SHACL VLDB 2024 — scales to 34M triples; removes reasoning-scale objection.
6. 🎯 xpSHACL (VLDB 2025) = the catala-explain analogue — LLM-backed violation reports + SHACLens visualisation + shacl-play documentation.
7. Legal-production SHACL thin but growing; INHERIT first cross-jurisdiction succession SHACL catalog = novel positioning.
8. VC 2.0 uses JSON Schema in credentialSchema (W3C-endorsed); SHACL layered on issuer side.
9. Phase-1 targets SHACL Core 1.0 + SHACL-AF 1.1 (not 1.2); dual-engine pySHACL 0.31 + TopBraid 1.5.

T5 SKOS + XKOS + OMG Commons Classifiers — 5 discoveries (MEDIUM-LOW)

1. SKOS frozen-stable — W3C Rec 2009 + no successor WG + no post-2020 errata. Good for ISO-track stability.
2. OMG Commons 1.3 beta (Dec 2025); Commons 1.2 adopted June 2025; cmns-* now supertype of ~90% of FIB-DM entities.
3. XKOS has no releases — rolling-spec model; must pin commit-SHA (new discipline gap).
4. SkoHub SHACL-Actions (Apr 2024) — production-ready GitHub Action for SKOS SHACL CI. Candidate A-21 gate.
5. Legal-domain SKOS ecosystem active — EuroVoc + ELI 3rd-ed 2024 + KEVLAR + FOLIO + AustroVoc + Liquid Legal Institute registry.

T6 PROV-O — 5 discoveries (MEDIUM-HIGH)

1. PROV-O core frozen — W3C Rec 2013-04-13 unchanged; 12-year stability = zero churn risk for A-22 FREE-tier.
2. 🎯 PROV-O → BFO total alignment paper (Prudhomme et al., Nature Sci Data 2025; arXiv 2408.03866; Zenodo 10.5281/zenodo.14692262) — every PROV-O class + property mapped to BFO-ISO + CCO + RO. Free BFO-ISO alignment for INHERIT.
3. nexus-prov (BlueBrain) SHACL shapes — drop-in validation saves ~200-500 LOC.
4. VC 2.0 + PROV unstandardised but tractable — VC as prov:Entity; issuance as prov:Activity; natural JSON-LD context union.
5. Catala-evaluation-as-prov:Activity = unoccupied territory — INHERIT first-mover publishable contribution.

T7 Akoma Ntoso + ELI + ECLI — 6 discoveries (MEDIUM-HIGH)

1. UK legislation.gov.uk /data.akn content-negotiation endpoint — entire UK corpus converted 2014.
2. Switzerland fedlex.admin.ch publishes both ELI URIs AND AKN XML since May 2022 — strongest Phase-1 Swiss precedent.
3. ELI 2024 releases: Ontology v1.5 + Impact v1.0 + DL v3.0 — ELI-DL adds in-flight/pending/repealed state semantics.
4. ECLI production via e-Justice Portal — 22+ EU member states + CJEU + ECtHR adoption.
5. catleg is fetcher/differ/skeleton-generator, NOT AKN pipeline stage — refines earlier assumption.
6. No AKN profile for Islamic/halakhic/Hindu — custom URN registry + SKOS mapping needed (T-NEW-URN-FAITH-01).

T8 ROBOT — 5 discoveries (MEDIUM-HIGH)

1. ROBOT v1.9.10 (Feb 2025); ODK 1.6.1 (Mar 2026) bundles ROBOT + SSSOM-Java 1.10.0 + ELK 0.6.0 + GitHub Actions templates.
2. ensure_owl2dl_profile: TRUE is ODK default — directly satisfies A-67 with 2-line CI step.
3. ROBOT has NO built-in SHACL — SHACL is SIBLING gate (pySHACL GitHub Action), not nested.
4. SSSOM-Java 1.10.0 sssom:inject ROBOT plugin — production-ready A-33 CI integration.
5. ROBOT + SSSOM + ODK = unified Phase-1 CI substrate; multi-day setup now hours.

T9 OpenAPI 3.1/3.2 + Arazzo — 5 discoveries (MEDIUM-HIGH)

1. OpenAPI 3.2.0 released 19 Sep 2024 (not future) — strictly-3.1-compatible + nested tags + SSE streaming + device-flow OAuth. 3.1.2 same day.
2. Arazzo 1.0.1 (Jan 2025) — 22 months mature; sole normative source; IANA-registered MIME types.
3. Arazzo-gated MCP is industry-consensus pattern — Kilcommins + Speakeasy + Zuplo + Nordic APIs 2024-26 converge. AutoMCP research (arXiv 2507.16044) 76→94.2% with repair.
4. A-82 18th-gate O3 hybrid recommended — Sourcemeta CLI + Speakeasy (oasdiff + Spectral + Arazzo validator + SDK contract test + MCP drift + JAIRF scorecard threshold).
5. Jentic JAIRF = external AI-readiness credential — 6-dimension scorecard; Erik Wilde Head of Enterprise Strategy Sep 2025. → T16 deep-dive.

T10 SSSOM — 6 discoveries (MEDIUM-HIGH)

1. SSSOM 1.0.0 shipped 9 Aug 2024; SSSOM 1.1 in active dev — pin 1.1 from day one (1.0 self-contradictory).
2. Dual mature toolchain — sssom-py (Python) + sssom-java 1.5.0 (May 2025) with ROBOT plugin.
3. ROBOT sssom:inject = canonical A-63 gate — declarative TSV → OWL materialisation.
4. Mondo Feb 2026 transition makes SSSOM authoritative in OBO Foundry.
5. SSSOM → SHACL auto-generation — eliminates T4/T10/T6 author-then-drift.
6. 🎯 First-mover positioning: no FIBO/CIDOC CRM/LADM/LKIF SSSOM catalogs exist. INHERIT Phase-1 would be first cross-domain legal/financial/heritage SSSOM catalog — ISO PAS positioning angle.

T11 F* + Mopsa — 5 discoveries (MEDIUM-HIGH)

1. F v2026.04.17* (18 April 2026) — Pulse separation-logic sublanguage + Low* removed. Z3 4.13.3 default since v2025.09.04.
2. F scope narrower than marketing*: translation-correctness only; NOT end-to-end verified compiler.
3. Rocq parallel Catala formalization (CatalaLang/catala-formalization) — 4,261 spec + 2,846 proof LOC + 527 commits. Natural insurance if F stalls*.
4. Mopsa SV-COMP 2026 Gold (SoftwareSystems) + Bronze (TrueOverall); Try-Mopsa browser-deployable (VMCAI 2026).
5. DGFiP + CNAF running Catala POCs — French government commitment; DGFiP targeting ~2028-2029 production.

T12 External ontologies (CIDOC CRM + FIBO + LADM + BFO) — 12 discoveries (MIXED per family)

CIDOC CRM (LOW — tier unchanged):

1. v7.1.3 (Feb 2024); ISO 21127:2023 status renewed; v7.2 in preview.
2. LRMoo bibliographic-descent pattern = candidate Phase-1.5 pattern for testamentary-instrument version chains (will → codicil).
3. Virtual Record Treasury of Ireland = closest legal-archive precedent beyond heritage.

FIBO (MODERATE-HIGH — tier raised): 4. FIBO 2025/Q4 Production (Jan 2026) = 3,173 entities. 5. OMG Commons 1.3 beta = ultimate supertype for ~90% FIB-DM entities. 6. 63 classes + 50 properties deprecated Q4 2025 (Commons 1.3 integration). 7. 🆕 5 new ultimate supertypes: Situation / Constituent / Designation / Collection / Aspect — Designation + Situation directly INHERIT-relevant. 8. NEW TENSION T-NEW-FIBO-CMNS-PIN (HIGH): Phase-1 must pin to Commons 1.3 stable, not beta/RTF.

LADM (MODERATE — scope-decision needed): 9. 5 of 6 parts published (1/3 in 2024; 2/4/5 in 2025) — far beyond 2012-baseline. 10. NEW TENSION T-NEW-LADM-SCOPE (MEDIUM): decide in-scope parts.

BFO (HIGH positive — tier reinforced): 11. BFO 2020 / ISO 21838-2:2021 stable; DoD + IC adopted Jan 2024 as baseline standard — strongest institutional backing of any candidate. 12. Nature Sci Data 2025 PROV-O↔BFO mapping paper validates Shape X iii peer-lock rationale.

T13 LinkML — 7 discoveries + library uplift (HIGH-conditional-on-spike-success)

1. LinkML directly resolves Viotti’s “constraints not modeling” admission — modeling layer + multi-target generation.
2. Production-mature + peer-reviewed: v1.10.0 (Feb 2026) + GigaScience 2026 paper (Moxon et al.; 36 authors; arXiv 2511.16935; PMC12993438).
3. Generator matrix 15+ targets: 8 STABLE+HIGH (JSON Schema 2020-12 + JSON-LD + SHACL Core + OWL + Pydantic v2 + Python dataclasses + Markdown + ER diagrams).
4. Biomedical adoption deep (Biolink + Monarch + NMDC + CZ CELLxGENE); non-biomedical thin (Gaia-X cited only).
5. Native SKOS enum integration — EnumDefinition maps 1:1 onto our 4 SKOS classifier schemes.
6. schema-automator reverse path — existing JSON Schema importable (imperfect but tractable).
7. Viotti/Sourcemeta likely COMPLEMENT not competitor — LinkML compiles TO JSON Schema; increases Blaze/Sourcemeta usage.

LinkML library uplift (2026-04-25 same-session):

• Rich acquired GigaScience 2026 PDF within same session
• Indexed at ~/off-github/library/indexed/moxon-linkml-framework/
• 11 ULID-cited quotes + 5 synthesis claims extracted
• Library confidence 30% → 75-80%
• Signal #1 (licence) RESOLVED: “a mixture of BSD-3 and Apache 2.0” — Apache-2.0-aspirational core compatible

Proposed spike: SPK-LINKML-01 — 6 weeks, 3 pilots (A-90 GoverningLaw → 4 SKOS classifiers → Credentials module); Week-3 go/no-go gate.

---

6 cross-technology convergent threads

Thread 1 — Upper-ontology stack consolidation (ISO-track strengthening)

PROV-O → BFO alignment paper (T6) + OMG Commons 1.3 as FIBO ultimate supertype (T5+T12) + BFO DoD/IC adoption + Shape X iii PROV-O+BFO peer-lock. INHERIT v2 has stronger ISO-track upper-ontology story than baselined.

Thread 2 — SHACL auto-generation pipeline (Phase-1 authoring efficiency)

SSSOM→SHACL auto-generation (T10) + nexus-prov SHACL shapes drop-in (T6) + SkoHub SHACL-Actions (T5) + pySHACL GitHub Action (T8) + ROBOT-sibling-SHACL-gate (T8) + LinkML SHACL Core generator (T13). Phase-1 SHACL authoring could be 50-70% auto-generated.

Thread 3 — ROBOT + SSSOM + ODK as unified Phase-1 CI substrate

ODK 1.6.1 (T8) + sssom-java 1.10.0 ROBOT plugin (T10) + ROBOT validate-profile default (T8) + ELK 0.6.0 default (T8). Phase-1 CI substrate is Docker-image-ready; multi-day setup → hours.

Thread 4 — Arazzo-gated MCP as AI-vendor commercial architecture

Arazzo 1.0 production (T9) + AutoMCP research (T9) + Jentic JAIRF external credential (T9→T16) + Speakeasy/Fern-Postman ecosystem. Phase-C Step 4 Commercial-model has concrete pattern: workflows as MCP tools, not raw endpoints; JAIRF scorecard as procurement gate.

Thread 5 — Catala verification stack robustness

F* v2026.04.17 active (T11) + Rocq parallel formalization as insurance (T11) + Mopsa SV-COMP 2026 gold (T11) + DGFiP/CNAF Catala POCs (T11). Catala adoption risk is substantially lower than scored during initial Catala spec.

Thread 6 — Unified auditable explanation layer (COMMERCIAL DIFFERENTIATOR)

catala-explain (Catala) + xpSHACL (SHACL violations) + ProvToolbox (PROV-O) + ProbateGrantCredential .docx (VC 2.0 issuance) + LinkML documentation generators. InheritKit unified commercial explanation artifact pipeline — THE differentiator for AI-vendors. Every legal decision + validation + credential + provenance produces lawyer-readable .docx.

Thread 7 (emerging with T13) — Chris Mungall intellectual-lineage unification

LinkML (T13) + SSSOM (T10) + OBO Foundry (T8 ROBOT) all share Chris Mungall stewardship + Nicolas Matentzoglu overlap. Adopting LinkML inherits tight integration with SSSOM + ROBOT/OBO Foundry workflows — not three separate technology choices; one intellectual lineage with unified governance.

---

All ~30 new tensions surfaced

Phase-C Step 2 Akoma Ntoso (from T7)

• T-NEW-URN-FAITH-01 (MEDIUM) — URN scheme registry for faith-tradition citations
• T-NEW-ELI-DL-01 (LOW) — ELI-DL v3.0 adoption for in-flight legislation
• T-NEW-CATLEG-PATTERN-01 (Phase-D) — INHERIT-equivalent fetcher/differ adapter

FIBO + external ontology pinning (from T12)

• T-NEW-FIBO-CMNS-PIN (HIGH) — pin to Commons 1.3 stable
• T-NEW-LADM-SCOPE (MEDIUM) — decide in-scope LADM parts

A-33 SSSOM catalog refinements (from T10)

• Pin sssom_version: 1.1 from day one
• Add sssom:inject CI step as A-99 candidate
• Extend SSSOM schema with inherit_tier + iri_verified_date columns

A-21 CI gate additions (from T5 + T8 + T11 + T4)

• SkoHub SHACL-Actions gate (T5)
• pySHACL GitHub Action sibling gate (T8)
• Mopsa date-rules gate on 7 named INHERIT date-rules (T11)
• 5 module-level Catala gates (T11)
• xpSHACL dry-run gate (T4)
• SHACL 1.2 forward-compat scan (T4)
• SHACL shape coverage report (T4)

Version-pinning discipline gaps (from T5 + T12)

• XKOS commit-SHA pin (T5)
• cmns-cls dated-URI pin (T5)

T13 LinkML spike-candidate amendments

• A-NEW-LINKML-SPIKE (authorise SPK-LINKML-01)
• A-NEW-LINKML-PIN v1.10.x (conditional)
• A-NEW-LINKML-META source-of-truth repo discipline (conditional)
• A-NEW-LINKML-ENUM SKOS classifier lock (conditional)
• A-NEW-LINKML-CATALA linkml2catala adapter

Phase-C Step 3 SSI spec (from T2)

• T-NEW-A43-REMAINING-4 — lock Attorneyship + IntermediaryAuthorization + Practitioner + Kinship VC placements
• T-NEW-EUDI-PHASE-1 — EUDI alignment as Phase-1 constraint (Swiss jurisdiction)
• T-NEW-KERI-PROMOTION — KERI from Phase-2+ to Phase-1-eligible
• T-NEW-CATALABACKED-CLAIM — CatalaBackedClaim pattern W3C VC CG contribution

T3 JSON Schema refinements

• T-NEW-JSONSCHEMA-V1-2026-PIN — pin Phase-1 to v1/2026 stable-evolving spec
• T-NEW-BLAZE-ADOPTION — Blaze as InheritKit SDK reference validator
• T-NEW-INHERIT-DIALECT-URI — publish https://schemas.inherit.org/dialect/v1
• T-NEW-LINKML-SPIKE (per T3 recommendation; also T13)

🚨 Urgent commercial

• T-NEW-SOURCEMETA-MCP-COMMERCIAL — Richard-action: raise two-tier architecture with Viotti BEFORE Sourcemeta One ships MCP

iri-verify tooling (from T12)

• Build iri-verify CLI in Python (rdflib + owlready2 + requests + JSON lockfile with SHA256)

Phase-1.5+ pattern candidates (from T7 + T12)

• T-NEW-CL-CIDOC-LRMOO (LOW) — LRMoo bibliographic-descent for testamentary-instrument version chains

T1 MCP opportunities

• T-NEW-MCP-TASKS-DOCX — use Tasks primitive (SEP-1686) for catala-explain .docx delivery
• T-NEW-MCP-SERVER-CARD — publish .well-known/mcp/server-card.json from day 1
• T-NEW-MCP-VC-AUTH-SEP — submit MCP-VC auth profile SEP to AAIF (standards contribution)

T4 SHACL open questions

• SHACL 1.2 Rules migration spike
• GraphDB hosting evaluation
• SHACL↔Catala bidirectional investigation
• SHACL Compact Syntax drop decision

---

Rich-action conversations identified (5)

Action	Rationale	Urgency
🚨 Juan Cruz Viotti (Sourcemeta)	T3 found Sourcemeta One MCP “coming soon” overlaps with InheritKit revenue thesis	URGENT (before Sourcemeta ships MCP)
Erik Wilde (Jentic)	JAIRF certification pathway for InheritKit AI-vendor procurement	HIGH
Chris Mungall (LinkML + SSSOM + OBO Foundry)	Unified toolchain adoption; community-lineage relationship	MEDIUM
Damien Goutte-Gattat (sssom-java maintainer)	SSSOM catalog integration + ROBOT plugin Phase-1 pilot	MEDIUM
Bruno Lowagie (iText)	Commercial-terms research for AGPL+commercial dual-license	MEDIUM

---

Strategic shifts identified

Biggest: ISO-track positioning STRONGER than baselined

BFO DoD adoption + PROV-O↔BFO alignment + FIBO cmns-* supertype + LADM 5-of-6 parts published = materially upgrades INHERIT v2’s JTC 1 PAS story.

Commercial reframing: explanation-as-a-service not data-licensing

catala-explain + xpSHACL + Thread 6 unified explanation layer + OSCAL (T14 pending) → 4-tier InheritKit commercial model (Free / Pro / Enterprise / AI-Vendor) with .docx explanation artifact per query + optional audit-pack.

Phase-1 authoring shift candidate (SPK-LINKML-01)

If spike succeeds: 40-60% authoring reduction via single-source LinkML authoring with multi-target generation.

Phase-1 CI substrate compression

ODK 1.6.1 bundles everything; multi-day setup → hours; Docker-image-ready.

Phase-1 SHACL auto-generation

50-70% of shapes auto-generated from SSSOM catalog + external-ontology imports.

Standards-contribution opportunities (reputational compound)

1. Catala-evaluation-as-prov:Activity (PROV WG contribution)
2. MCP-VC auth profile SEP (AAIF contribution)
3. LinkML legal-domain extension (Biolink-analog for legal)
4. First cross-domain legal SSSOM catalog (OBO Foundry precedent)
5. First cross-jurisdiction succession SHACL catalog

---

Zero retroactive impact on locked decisions

All Phase-B locks hold:

• Wills v1.3 (9 amendments) ✅
• Trusts v1.1 (3 amendments) ✅
• Probate v1.1 (4 amendments) ✅

Catala scope-δ lock holds (9 components; strengthened by T11 Rocq formalization + T13 LinkML precedent).

A-22 iv-3 alignment matrix tier assignments all hold (FIBO internally strengthened; CIDOC CRM SPARINGLY correct).

A-21 CI chain at 18 gates (extensions queued for Phase-1 authoring only).

---

Library ingestion during programme (2026-04-25 same-session uplift)

moxon-linkml-framework (GigaScience 2026; PMC12993438; DOI 10.1093/gigascience/giaf152)

• Peer-reviewed primary source; 13 pages; 36 authors
• 11 ULID-cited quotes + 5 synthesis claims extracted
• T13 library-grounding 30% → 75-80%
• T13 Signal #1 resolved favourably: BSD-3 + Apache-2.0 licence mixture
• Added to library-catalogue.md Tier-1

Moxon notes directly support:

• Viotti’s “constraints not modeling” peer-review-confirmed
• Single-source-of-truth claim peer-review-stated
• Generator matrix confirmed (JSON Schema + SQL + RDF + OWL + SHACL + Python)
• Honest author-stated conversion limitation (validates SPK Week-3 fidelity gate)
• Non-biomedical precedent: Gaia-X
• LLM integration precedent: OntoGPT
• LinkML-Map extends SSSOM to structural-schema-level (unified lineage)

---

Round 2 dispatched 2026-04-25 (T14-T19 pending return)

Six new subagents dispatched following T1-T13 programme completion + Rich’s “before returning to planning” research extension:

Tier A — Commercial-strategy critical (3):

• T14 OSCAL (NIST audit data-model) — Phase-C Step 4 AI-Vendor tier audit-pack
• T15 xpSHACL + SHACLens + shacl-play — Thread 6 unified explanation layer depth
• T16 Jentic JAIRF — AI-vendor procurement certification pathway

Tier B — Integration-architecture critical (3):

• T17 OntoGPT — LinkML + LLM + MCP tool-input schema pattern
• T18 Biolink Model — LinkML flagship production deployment lessons
• T19 Gaia-X — Non-biomedical LinkML production precedent + EU federation

---

Programme statistics

Metric	Count
T-files created	19 (T1-T19)
Subagents complete	13 (T1-T13)
Subagents pending	6 (T14-T19 Round 2)
Top-tier discoveries	~50 across T1-T13
New tensions surfaced	~30
Cross-technology threads	6 (Thread 7 emerging)
A-NEW amendment candidates	~20
Rich-action conversations	5
Library items indexed this session	1 (moxon-linkml-framework)
Retroactive disruption to locks	0
Strategic shifts identified	5
Standards-contribution opportunities	5

Programme cost so far: ~£50-75 subagent + Claude in-session work. Round 2 adds ~£30-40.

Programme yield: Phase-C Step 2/3/4 spec authoring research-grounded baseline + Phase-1 CI substrate compression + Phase-1 SHACL auto-generation + Phase-1.5 spike candidate + commercial-model reframing + ISO-track positioning uplift.

Related memories

• project_technology_research_programme_2026_04_25 — programme plan + findings summary
• project_technology_research_synthesis_2026_04_25 — top 10 synthesis + action plan
• project_catala_review_plan_2026_04_25 — 14 Catala improvements
• project_catala_ecosystem_maturity_2026_04_25 — 3 Catala discoveries (precedent)
• project_catala_explain_as_commercial_artifact — commercial reframing
• feedback_library_web_research_narrow_margins — 11th auto-deep-dive pattern
• project_ai_vendor_commercial_model — updated with explanation-artifact reframing

CHANGELOG

• 2026-04-25 — Comprehensive discoveries catalogue created at Rich’s explicit request (“please save details of all the discoveries we have made”) to preserve everything before Round 2 T14-T19 subagents return + before returning to Phase-1 planning. Durable record surviving context compaction.