LATTICE Developer Coordination Index

Last updated: 2026-09-25 — ADR-A97 makes MORK OWL 2 DL (Mork 0.4.0). ADR-A83 reasoning harness delivered (eligibility-compiler Part B). applied-ontology-readiness AOR-10, AOR-11 and AOR-14 to AOR-17 implemented, ADRs A-83, A-86, A-93 to A-96 accepted. documentation-link-repair planned
Purpose: Single source of truth for all active units of work: sketches, plans, status, reviews, and validation packs.
Format: By unit identifier, with links to all related documents and current status.

How to read this index

Each work unit has:

• Status (✅ complete, 🚧 in-progress, ⏳ planned, 🗄️ archived)
• Unit ID (referenced in reviews and handoffs)
• Documents (sketch, plan, status, review, validation pack)
• Blockers (if any)
• Key findings (major discoveries or decisions)

Navigation: Use Ctrl/Cmd+F to search by topic (e.g., “RDF”, “MTP”, “housekeeping”) or by unit ID.

Part I — Completed Work

Persistence and IRI patterns: the change package at a glance

Is it complete? Almost. The documentation is complete and every review is closed. The compiler is five slices into a six-slice re-sync, and the sixth (documentation close-out) is in progress this session. Housekeeping has not started, and the identity decisions await ratification. Row by row (as of 2026-09-25):

1. RDF/SPARQL Implementation Patterns and Persistence Compiler

Slice Completion Status

| Slice | Deliverable | Status | |——-|————|——–| | 1 | rdf-sparql-patterns-guide.md — consolidated 4 source notes into authoritative reference (30 chapters covering K/O/C/T/QP patterns) | ✅ Complete | | 2 | ontology/persistence + tools/persistence compiler + policy enforcement + doc deltas | ✅ Complete | | 3 | platform/housekeeping module (scaffolding, not execution) | 🚧 Planned, not started |

Key Findings

• Target model discovery: Classes need paired (class, deployment) targets within each graph scope, added dal:coversClass property
• SPARQL validity bugs fixed: Property paths invalid in DELETE/INSERT blocks; payload triples need a text slot instead of SPARQL variables (originally #PAYLOAD#, since persistence-compiler-iri-sync Slice 2 the Mustache slot })
• Mustache parsing gotcha: Comments cannot contain bare }} without breaking parsing. Extends to a full } tag quoted as prose inside a `` comment (found in persistence-compiler-iri-sync Slice 5’s key-claim-merge-rewrite.mustache) — describe a slot in words in template comments, never quote its Mustache syntax.
• pyshacl semantics: allow_warnings=True needed for non-blocking sh:Warning severity

Blocks

• Housekeeping module first cut (Slice 3, depends on Slice 2 compiler)

Epic decomposition into phase plans is no longer blocked: the required Phase 2 (P2.1/P2.3/P2.4) revision integrating this compiler is complete — see Part II, §6.

1a. Persistence Compiler / IRI-Patterns Sync (new, tracks the 2026-09-23 gap)

Gap summary

Three new profile dimensions (dal:IdentityProfile, dal:EpochProfile, dal:PrivacyProfile), extensions to six existing ones, and eight new SHACL shapes were added to ontology/persistence, none consumed by tools/persistence. One finding (G1) is a live correctness issue, not just missing coverage: the compiler’s CAS/tombstone templates generate the epoch-guard shape the vocabulary now documents as unsafe (dal:RowLevelGuardOnly), unconditionally, with no way to configure the safe dal:DatasetLevelGuard alternative.

Slices

1. ✅ Complete, 290/290 passing. Dataset-level epoch guard — dal:epochGuardScope resolvable, dal:DatasetLevelGuard template variant for the three named write shapes, warning diagnostic fires even on the platform baseline default. See the VP.
2. ✅ Complete, 526/526 passing. Extension properties resolved one dimension each, baseline defaults, two refusals and six warnings mirroring the SHACL shapes, dal:PreCreatedRow emits bootstrap-version-row, request-time values as Mustache slots. See the VP.
3. ✅ Complete, 570/570 passing. Identity resolved per resource role (identity:<Role>), winning profile node as a unit, emitted to the compiled profile, five refusals and one warning. See the VP.
4. ✅ Complete, human-validated 2026-09-25 (669 passed, adversarial probes checked). Privacy/erasure profile (dal:privacyClass/dal:erasureStrategy/dal:erasurePrecedence/dal:perSubjectScoped, one dimension each) plus the G3 remainder (dal:epochAuthority promoted to its own dimension, carrying the remaining restore-surface properties as its extras), two refusals mirroring dal:PersonalDataRequiresErasureShape and dal:PersonalDataReceiptCompatibilityShape, and Worked example 4’s privacy profile now resolving and emitting cleanly. See the VP.
5. ✅ Complete, human-validated 2026-09-25 (774 passed). Uniqueness dal:onViolation selects a reconciler operation, never the guarded write itself (decision 1, Option A — guide §7.5, not §6): key-claim-duplicate-audit (dal:Reject, the default), key-claim-merge-rewrite (dal:Merge, with MergeRelationRequired), key-claim-quarantine (dal:Quarantine). dal:ClaimScheme dal:Dual rotation selects key-claim-write-dual.mustache. The registry-token digest-scheme relaxation (found in identity-minting M3) is also folded in: dal:DigestSchemeRequiredShape now exempts dal:PositionDerivedEvent + dal:RegistryTokenDerivation only. ontology/persistence bumped PATCH, 0.2.0 → 0.2.1. One issue found by the run itself, fixed: a mustache-comment quoted its own tag syntax as prose, which this compiler’s renderer does not treat as inert. See the VP.
6. 🚧 Documentation close-out, in progress this session

1b. IRI and RDF Patterns, Post-3866b21 Remediation

All 7 critical (A), 14 major (B), 9 safety (C), 7 cross-document (D) and 10 editorial (E) findings are fixed in the documents. The main design choices (lazy epoch rebase, request digest, pinned heads, row-driven gap audit, NFKC_Casefold) are recorded once in the status record. The compiler templates now generate the corrected write shapes and audits. A pre-existing bug that wrote key claims into the txn graph is fixed.

2. LLM Training and MORK Teaching Pack (MTP) Generation

Completion Status

• ✅ Facts extraction (S1)
• ✅ Doctrine pinning (S2)
• ✅ Axiom ownership (S3-S5)
• ✅ Lensing (S6)
• ✅ Mutation-derived ground truth (S7-S9)
• ✅ Cassette validation (S10)
• ✅ Rendering (S11-S13)
• ✅ Test determinism (L2), roundtrip validation, mutation checks

Blocks

• Phase 7 backend integration (runtime API for MTP consumption)

3. Repository Topology and Documentation Governance

Deliverables

• ✅ Folder structure relocated per ADR-A77
• ✅ mise remains sole orchestration entry point
• ✅ Epic Decomposition Model documented in copilot-instructions
• ✅ Validation Pack and traceability model specified

4. MORK Eligibility Compiler

Deliberate Non-Coverage

• No native backend
• No Executable Projection Contract layer
• No profile-level aggregate artifact
• SWRL positive-only per ADR-A24

Blocks

• Interval SWRL rules use swrlb builtins, which HermiT does not evaluate. They wait on an Openllet adapter (AGPL-3.0, usable only behind the ADR-A83 isolation, plan §B.4).

5. Governance Surfaces (MORK Review Workbench, Surface Studio)

Deliverables

• ✅ Review Workbench: 6 decision buttons (Confirm/Retarget/Reshape/Decline/Teach/Defer), evidence ribbon (Witness/Coverage/Alternative)
• ✅ Studio: 4 fixture contracts with lifecycle transitions and diff viewer
• ⏳ Role model demonstrated (enforcement not wired)
• ⏳ Token ribbon partial (needs MCN source view)

Remaining Work (3 slices per plan)

1. Phase 2 Slice: Studio backend wiring (MORK mapping integration)
2. Phase 5 Slice: Bench permission model + enforcement
3. Phase 5 Slice: Evidence projection state

Part II — Active/Planned Work

6. Epic Decomposition: Lattice Platform Agentic Development

Scope

• 16 tracks (T-DEC through T-SEC) with hard orderings
• 9 milestones (M0: walking skeleton through M9)
• 25+ modules in Part 1
• Phases 0-1 at slice granularity; Phases 2-4 lighter initially

Required Epic Plan Revision — ✅ Done (2026-09-22)

Part 6 (Phase 2) of the epic plan now integrates the persistence compiler instead of constructing SPARQL ad hoc:

• P2.1 (C-05, mapping plan compiler): new slice P2.1.4a resolves each writeTarget’s ontology/persistence profile and emits a dal:CompiledProfile plus instantiated (still-unbound) SPARQL Update text via tools/persistence, as part of CompiledMappingSet.
• P2.3 (C-04, ingestion gateway): the assert step (P2.3.1) and commit step (P2.3.6) bind request-scoped parameters into those pre-instantiated templates — never construct SPARQL at request time. Deliberately scoped narrower than the still-deferred Request Query Mapping library (Part 13, open question 11).
• P2.4 (C-12, query plane): the typed read API (P2.4.1) reuses the same resolved profile (ordering grain, receipt model) to shape read queries, consistent with the write path by construction.

See rdf-sparql-patterns-phase-plan.md and persistence-profile-substrate.md for the underlying substrate; no new ADR was needed (refinement of ADR-A78/A79 within existing guardrail G5).

Optional Phase 0.2–0.4 Integration — ✅ Done (2026-09-22)

Per COORDINATION_REORG_HANDOFF.md, Phase 0 of the epic plan (Part 4) now carries a “Persistence-pattern coherence” note plus cross-references so its vocabulary is dal:-nameable from day one, without Phase 0 depending on tools/persistence:

• Pattern C (CAS) ↔ dal:ConcurrencyStrategy, cross-referenced at P0.5.2 (conditionalWrite/guardSatisfied) and P0.5.6 (concurrency TCK).
• Pattern T (named-graph-per-batch) ↔ dal:MetaTopologyProfile/dal:ReceiptModel, cross-referenced at P0.1.5 (ADR-A65) and P0.3.6 (provenance.ttl).
• Pattern K (uniqueness) ↔ dal:UniquenessConstraint, cross-referenced at P0.1.3 through the configurable minting and identity profiles described by ADR-A82.
• Pattern O (dense ordering) ↔ dal:OrderingGrain/dal:DatasetTierModel, cross-referenced at P0.4.4 (canonicalisation) and P0.5.7 (CommitSequence + ordering TCK).
• New slice P0.3.9 (documentation only) produces ontology/persistence/docs/platform-vocabulary-alignment.md, proving no namespace/semantic collision between the six dal: dimensions and the platform vocabulary P0.3 defines.
• New guardrail G11 (§0.4): no dal:-governed write path may hand-construct SPARQL, reserved at P0.2.7 and enforced in full once P0.5.2’s write surface exists.

Epic Decomposition — ✅ Done (2026-09-22)

Each phase plan adds three obligations the epic’s slice tables name but never tabulate: a docs/architecture deliverables table, a subproject-README table, and a solution-design-specification.md delta table (which SDS section changes, and at which slice).

Three mechanical corrections made to the epic during decomposition (not architectural decisions — corrections against already-established convention): deploy/ → deployment/ (root already exists under that name), make/just → mise tasks (copilot-instructions mandates mise as sole orchestrator, ADR-A29 not superseded), docs/adr/ → docs/architecture/decisions/.

Pre-execution decision points surfaced, not resolved (see phase-0-plan.md §2):

• Relationship between the epic’s proposed platform/graph-spi family and the existing platform/semantic-dataset-spi/semantic-dataset-fuseki — blocks P0.5.1.
• apps/surface-studio/apps/mork-bench (epic’s Part 1 names) vs existing apps/surface-contract-studio/apps/mork-review-workbench — blocks P1.10, lower urgency.

No code was written and no new Phase 0→Phase 2 dependency was introduced — this is cross-referencing and one new documentation-only slice.

IRI and Identity Patterns — Proposed (2026-09-23)

The unratified universal IRI policy in ADR-A51 was superseded by ADR-A82. The new IRI and Identity Patterns guide treats entity, aggregate, component, lineage, content-revision, graph-locator, key-claim, and event-occurrence identities as independently configurable patterns. It crosswalks those choices to the K/O/C/T/QP RDF and SPARQL patterns and specifies the future dal: identity-profile vocabulary boundary. The dal:IdentityProfile vocabulary is specified in ontology/persistence, and the compiler resolves it per resource role (persistence-compiler-iri-sync Slice 3, complete). The guide and ADR-A82 were revised on 2026-09-23 by iri-patterns-post-3866b21-remediation and await ratification (P0.1.3). iri-policy.md is now a short historical record.

7. Surface Projection and MORK Unified Compiler (Phases 0–10)

Phase Completion Status

Key Findings

• Projection mechanism: Unified abstraction for promotion (restating as direct properties) and shadow indexing (generating lookup symbols)
• Deterministic lowering: Surface contracts lower to MORK with full provenance tracking and minimal-scope regeneration
• Compiler family: Four production backends (SPARQL, SHACL, SWRL, Eligibility IR) + one deferred native backend
• Parity guarantees: Generated output matches source contract semantics at ≥99% fidelity
• Verification checklist: 7/8 items passing; SWRL reasoner integration (Phase 8, item 7) remains open

Completed Slices

Operational Documentation

• Invalidation Runbook — 4 operational scenarios with release gates
• Revision Lifecycle — State machine, versioning, canonicalisation cutover procedures

Outstanding Items

| Item | Status | Link | |——|——–|——| | Signature scope and law X6 | ✅ Closed | Outstanding items §3.1 | | Foundation migration (srf:DerivedArtefact) | ✅ Done (ADR-A92) | Outstanding items §3.2 | | Profile identity assertion | ⏳ Blocked | Outstanding items §3.3 | | MORK toolchain join assumptions | ⏳ Needed | Outstanding items §3.4 |

Deferred Items (Explicit Scoping)

• srf:RangePartitionPopulation (bucketing) — blocked on Quantification layer
• Stacking beyond depth 1 — composition laws drafted, full support deferred
• Other eligibility strategies beyond IntervalContainment — deferred to broader Eligibility expansion
• srf:ExternalIndex — admitted but deferred, no criteria yet
• Entailment regimes beyond NoEntailment — deferred to Phase TBD

Blockers

• Phase 8 SWRL verification: Reasoner integration for generated SWRL rules (1/8 checklist item remaining)
• Phase 9 decomposition: Awaits Phase 8 sign-off

8. Housekeeping First Cut

Deliverables (planned)

• platform/housekeeping module scaffolding (not execution)
• Contracts and configuration model
• Generated queries per ADR-A80: the row-driven gap scan, the txn-cardinality fork audit and the two receipt-side duplicate audits
• The retention job’s own write queries: advance the low-water mark, carry live heads into the pinned-head graph, prefix-only drops (guide §24.2). See the plan’s Slice 3 “Changes since this slice was scoped” note

Blockers

• None. The compiler dependency is met

Vocabulary Scoped and Temporal Binding — Closed (2026-09-25)

The conformance package (12 fixtures, populated Vocabulary SHACL files, a deterministic reference resolver) is authored and verified: mise run check:vocabulary passes 14/14. Both real consumers now resolve scoped bindings instead of reading voc:boundScheme directly: Surface’s compiler calls the resolver in enumerate_population, and Eligibility’s HierarchyWellFoundednessShape checks every scheme a contract could resolve to. python -m unittest surface.test_surface -q passes 62/62 (including a new scoped-binding fixture and test), and mise run check:python-root passes 77/77 plus Phase 8 conformance. See the status record for exact findings.

Slices

1. ADR-A85, architecture mirror, fixture and traceability skeleton — done
2. Examples plus structural and SHACL-SPARQL validation — done, verified
3. Reference resolver and deterministic test suite — done, verified
4. Consumer/provenance checks and documentation close-out — done, verified (Surface and Eligibility both resolve scoped bindings)

Closure cross-reference (2026-09-25)

A cross-reference of this unit’s Surface/Eligibility integration against docs/architecture/rdf-sparql-patterns-guide.md found the work internally consistent, with four follow-on hardening items and one accepted terminology note. None block closure. See Vocabulary Consumer Hardening below.

8.5 Vocabulary Conformance — Closed

See the unit record above. This heading is intentionally a navigation anchor between active platform work and archived material.

Ontology Semantic Versioning — Implemented, ADR pending ratification (2026-09-25)

Every owl:Ontology document under ontology/ carried an ungoverned owl:versionIRI, or none, with no documented rule for when it changed. The sketch’s inventory found four distinct problems: no bump rule beyond one ad hoc note in vocabulary/README.md; an already-inconsistent versioning unit (spec/*.ttl vs vocab/*.ttl independently versioned and already drifted); two ontologies (MORK, SPC) with no version identity at all; and one (applied/insurance/contract.ttl) with two disagreeing version signals and a namespace base outside the lattice/ tree. ADR-A86 adopted SemVer 2.0.0 per owl:Ontology document, a MAJOR/MINOR/PATCH mapping table, a one-time baseline reset to 0.2.0, and a narrow “changed but not bumped” check.

Slices

1. ADR-A86, sketch, plan, status record, ADR-index and INDEX.md entries — done
2. Developer/agent guidance (ontology-versioning-policy.md + CONTRIBUTING.md/ontology-architecture.md cross-references) — done
3. Baseline reset (every in-scope document to 0.2.0, lattice/ base-URI normalisation, full owl:imports cascade, one discovered non-ontology importer fixed: tools/surface’s SURFACE_ONTOLOGY constant) — done, with one recorded deviation (README⇄spec extraction was not re-run; see status record)
4. Narrow “changed but not bumped” enforcement tooling (tools/ontology_version_check.py, mise run check:ontology-versioning) — done, validated by mutation probe

Discovered while implementing, not yet acted on: tools/literate_extract.py --check does not currently pass for any of the seven core literate-spec layers (pre-existing README⇄spec drift, unrelated to this unit). Recorded in the policy document as a candidate future remediation unit.

8.6 Ontology Semantic Versioning — Implemented, ADR pending ratification

See the unit record above. This heading is intentionally a navigation anchor between active platform work and archived material.

Vocabulary Consumer Hardening — Implemented (2026-09-25)

Four items carried forward from the vocabulary-temporal-binding closure review, all implemented:

1. Finding 3 (docs). vvp:resolvedAt clarified as a valid-time “as-of” point, not a transaction-time “recorded when” timestamp, in ontology/vocabulary/shapes/constraints.ttl, tools/vocabulary/README.md, and the validation pack.
2. Finding 4 (test). tools/vocabulary/tests/test_architecture.py: no module under tools/vocabulary/src/vocabulary/ may call a wall-clock function, mirroring tools/persistence. Mutation-probed in this session (fails when a datetime.now() call is introduced, passes otherwise).
3. Finding 1, Option B (compiler + CLI + law text). tools/surface’s CLI (command_compile/command_check/command_parity/command_mork) now refuses to default produced_at to wall-clock time for any contract whose population could resolve a caller-scoped voc:SchemeBinding (contracts_needing_explicit_resolution_time in cli.py); srf:R1’s law text and compile.py’s module docstring updated to state the determinism guarantee precisely. The parity command’s --shared-corpus path is not guarded (deliberate non-coverage; no scoped fixture is in that corpus today).
4. Finding 2 (ontology + compiler + manifest). New srf:resolvedAt, srf:resolvedBindingScope, srf:resolvedBinding, srf:resolvedViaFallback on srf:ReadSetEntry, populated on every BoundSchemeSource entry via a vocabulary.Resolution object now threaded through enumerate_population instead of discarded; a new srf:BoundSchemeSourceResolutionRecordedShape SHACL-SPARQL check; new tests proving the recorded trace changes when the caller’s context does.

Finding 5 (terminology overload of “scope”) remains accepted, no action.

Applied Ontology Readiness — all slices implemented (2026-09-25)

Thirteen gaps (AO1 to AO13) that any applied ontology meets: Eligibility examples that warn or violate since 9a12da4, no reproducible way to load LATTICE’s import closure, versioning guarantees not enforced in CI, compilers limited to interval conditions and to candidates held on questions, no OWL backend for design-time checks, provenance not aligned with PROV-O, and four substrate extensions (derived rates, calendar binding, per-unit alternative bounds, one obligation in several provisions).

Slices

1. Phase A, consumable baseline: AOR-1 governance records — done. AOR-2 examples and the declaration warning, AOR-3 versioning guarantees, AOR-4 import catalog — implemented, VPs
2. Phase B, executable coverage: AOR-5 to AOR-9 (concept conditions, hierarchical match, SHACL and SWRL, diagnostics, profile aggregation, conformance corpus, evidence bindings) — implemented. AOR-10 and AOR-11 (OWL backend and its checks) — implemented, VP
3. Phase C, substrate extensions: AOR-12 (Foundation derived-artefact contract) and AOR-13 (Executable to PROV-O) — implemented. AOR-14 to AOR-16 (one Quantification change) and AOR-17 — implemented, VPs

Documentation Link Repair — Pending (2026-09-25)

mise run topology:links reports 419 broken links (340 distinct). 290 are in the ignored Jekyll output docs/_site/, which the checker should not scan. The other 50 are wrong relative depths, moved targets, and targets that no longer exist. Five slices, about 180k tokens.

8.7 Applied Insurance Reference (Epic) — Proposed

Reference implementation of the applied insurance domain: module split, reference peril vocabulary, Eligibility over flat lists and several values with reviewed crosswalks for flat or taxonomic peril lists, exposure ontology, term parameters with derived liability direction, and optional compilation into Capacity. Seven substrate items (L-P1 to L-P5, spatial guidance, evidence resolution) run as a non-blocking track.

Part III — Archived/Historical Work

8. Phase Handoff Documents (0-6)

Reference: platform-continuation.md describes current platform state post-Phase-6.

9. Other Completed Units

Part IV — How to Navigate This Index

By Topic

Surface Projection and MORK Compiler

• Sketch: surface-projection.md
• Plan: surface-mork-unified-projection-plan.md
• Status: surface-mork-unified-projection.md, surface-outstanding-items.md
• ADRs: A16–A28 (13 decisions)
• Implementation: tools/surface/ (compiler), tools/mork_compilers/ (backends), ontology/surface/ (vocab + shapes)
• Tests: 61/61 Surface, 15/15 MORK backends, 346/346 LLM/MTP
• Operations: surface-invalidation-runbook.md, surface-revision-lifecycle.md

RDF/SPARQL & Persistence

• Guide: rdf-sparql-patterns-guide.md
• ADRs: A78, A79, A80
• Slice 2 (complete): ontology/persistence + tools/persistence
• Slice 3 (planned): Housekeeping

Vocabulary Binding

• Sketch: vocabulary-temporal-binding.md
• Plan: vocabulary-temporal-fixes.md
• Status: vocabulary-temporal-binding.md
• Validation pack: vocabulary-temporal-binding.md
• Implementation: ontology/vocabulary/shapes/, ontology/vocabulary/examples/ (12 fixtures), tools/vocabulary/ (reference resolver + pytest suite) — verified, 14/14 tests passing
• ADR: A85, Accepted
• Follow-on: temporal-binding-consumer-hardening.md — implemented (2026-09-25), see its own INDEX entry

Ontology Semantic Versioning

• Sketch: ontology-semantic-versioning.md
• Plan: ontology-semantic-versioning.md
• Status: ontology-semantic-versioning.md
• Policy: ontology-versioning-policy.md
• ADR: A86, Proposed

Applied Ontology Readiness

• Sketch: applied-ontology-readiness.md
• Plan: applied-ontology-readiness.md
• Status: applied-ontology-readiness.md
• ADRs: A87 to A92 and the A86 addendum, Proposed
• Implementation: docs/architecture/ontology-versioning-policy.md, tools/ontology_version_check.py (mise run check:ontology-versioning), every in-scope .ttl document reset to 0.2.0 — done, ADR ratification still pending

MTP & LLM Training

• Status: llm-training-mtp.md
• Execution: mtp-execution.md
• 346 tests passing

Architecture Decisions

• All ADRs: docs/architecture/decisions/
• Traceability: Referenced in each unit’s plan/status/validation-pack

Sketches & Plans

• Previous batch (llm-training, governance, eligibility): DESIGN_SKETCHES_INDEX.md
• Current batch (persistence): persistence-profile-substrate.md + rdf-sparql-patterns-phase-plan.md
• Epic: lattice-platform-agentic-development-v0.2.md

Validation Packs

• Location: docs/developer/validation/<unit-id>.md
• Most recent: persistence-substrate-and-compiler.md

By Phase

• Phase 0: Awaiting decomposition; blocks on RDF patterns integration
• Phase 2: Governance surfaces backend integration (3 slices planned)
• Phase 7: MTP backend integration (runtime API for curriculum delivery)

By Status

✅ Complete (Ready for Handoff or Integration)

1. RDF/SPARQL patterns guide (Slice 1)
2. Persistence compiler (Slice 2) — 774 tests passing after persistence-compiler-iri-sync’s Slices 1–5 (complete, human-validated 2026-09-25); that unit’s Slice 6 (documentation close-out) is its own final pass
3. LLM training / MTP generation — 346 tests passing
4. Repository topology (ADR-A77)
5. MORK eligibility compiler (awaiting runtime validation, not handoff)
6. Surface Projection Phases 0–8 — 61/61 Surface tests, 15/15 MORK backends, 7/8 verification items
7. Phase 0-6 handoff documents

🚧 In Progress

1. Epic decomposition (Phase 0-9 plans)
2. Persistence compiler / IRI-patterns sync (persistence-compiler-iri-sync) — Slices 1–5 of 6 done (774/774 passing, human-validated); Slice 6 (documentation close-out) in progress
3. Housekeeping first cut (Slice 3, scoped but not started)
4. Surface MORK Phase 8 verification (SWRL reasoner integration, 1/8 items pending)

⏳ Planned

1. Housekeeping execution (tied to future store SPI)
2. Governance surfaces backend integration (Phase 2/5)
3. Phase 0-9 implementation
4. Surface MORK Phase 9 — Migration guides and phased rollout
5. Surface MORK Phase 10 — Scale and performance optimization
6. Vocabulary scoped and temporal binding conformance — closed 2026-09-25
7. Vocabulary consumer hardening (temporal-binding-consumer-hardening) — implemented, not yet executed in this sandbox
8. Ontology semantic versioning (ADR-A86) — implemented, ADR accepted 2026-09-25
9. Applied ontology readiness (applied-ontology-readiness) — AOR-2 to AOR-9, AOR-3b, AOR-12/13 committed, the rest implemented, review requested
10. Documentation link repair (documentation-link-repair) — pending

🗄️ Archived

1. Phase 0-6 handoff documents (refer to individual status records)
2. Implementation handover from earlier phases

Part V — Key Decision Points (Awaiting Ratification or Decomposition)

Part VI — Traceability and Validation

Validation Packs (VPACKS)

• persistence-substrate-and-compiler.md — RDF patterns Slice 2
• persistence-compiler-iri-sync-slice-1.md — compiler sync Slice 1
• persistence-compiler-iri-sync-slice-2.md — compiler sync Slice 2
• persistence-compiler-iri-sync-slice-3.md — compiler sync Slice 3
• persistence-compiler-iri-sync-slice-4.md — compiler sync Slice 4
• persistence-compiler-iri-sync-slice-5.md — compiler sync Slice 5
• applied-ontology-readiness-aor-2.md to aor-9, aor-3b, aor-12, aor-13 — applied ontology readiness
• More to be created as each slice/phase completes

Test Taxonomy (L0–L8)

• L0: Build smoke
• L1: Unit / pure-function
• L2: Property & determinism
• L3: Contract validation
• L4: Component integration
• L5: System E2E
• L6: UI E2E
• L7: Non-functional (SLO, benchmark)
• L8: Hostile / security

Test Coverage Status

| Unit | Tests | Command | |——|——-|———| | Surface compiler | 61 passing | pytest tools/surface/src/surface/test_surface.py -v | | MORK compiler backends | 74 passing (2026-09-25) | mise run check:mork-compilers | | LLM/MTP | 346 passing | mise run check:mtp | | Persistence compiler | 774 passing (confirmed 2026-09-25, after Slice 5) | mise run check:persistence | | Ontology tools (catalog, versioning, Eligibility examples, PROV-O alignment) | 35 passing | mise run check:ontology-catalog | | Ontology versioning | tool run | mise run check:ontology-versioning |

Part VII — Open Questions (Blocking or Deferred)

1. Store SPI design — How does runtime connect compiled templates to live backend? (Deferred to separate SPI phase. The caller contract it must honour is now documented in tools/persistence/README.md, “Using the generated SPARQL directly”)
2. Phase 0.2–0.4 walking-skeleton pattern integration — done (2026-09-22): Part 4 of the epic plan carries the Pattern C/T/K/O ↔ dal: cross-references, new slice P0.3.9, and guardrail G11. See Part II, §6 and Part V.
3. MTP backend integration — What HTTP API shape for LLM curriculum consumption? (Phase 7, post-Phase-6)
4. Housekeeping execution — When is the housekeeping component itself executed (real-time vs. batch)? (ADR-A80 defers to future phase)
5. Surface MORK Phase 9 decomposition — How to slice migration guides and phased rollout? (Blocking Phase 9, needs decomposition after Phase 8 SWRL verification)
6. Foundation migration for srf:DerivedArtefact — resolved 2026-09-25 by ADR-A92 (AOR-12). Original question: Should profile identity be a Foundation concept or Surface-only? (Blocking profile identity assertion in Surface; see surface-outstanding-items.md). Proposed resolution: ADR-A92
7. MORK toolchain join assumptions — Pre-production verification checklist for join operations in mapping compilation (see surface-outstanding-items.md)

Document Maintenance

This index is updated when:

• A new sketch is commissioned
• A plan is created or revised
• A status record is completed
• A validation pack is accepted
• A unit transitions between status states (🚧 → ✅, etc.)

Last updated: 2026-09-25 — persistence-compiler-iri-sync Slices 4 and 5 human-validated (774/774 passing); Slice 6 documentation close-out in progress Last reviewed: 2026-09-22
Next review: Upon Phase 8 SWRL verification completion and Phase 9 decomposition