dot-iu-cutter v0.5 — evidenced_by Vocab+Cap-4 Design Report (design-only, route GPT/User)
dot-iu-cutter v0.5 — evidenced_by Vocab + Cap-4 Extension Design Report
Phase:
v0_5_evidenced_by_vocab_amend_and_cap4_extension_design· Nature:design_only· Date: 2026-05-18 Package (4 files, this is #4):
dot-iu-cutter-v0.5-evidenced-by-vocab-amend-design-2026-05-18.mddot-iu-cutter-v0.5-evidenced-by-cap4-checker-extension-design-2026-05-18.mddot-iu-cutter-v0.5-evidenced-by-verification-and-rollback-plan-2026-05-18.mddot-iu-cutter-v0.5-evidenced-by-vocab-cap4-design-report-2026-05-18.md(this file)
1. Executive summary
deliverable: design-only conversion of the APPROVED evidenced_by APR into concrete
vocab-amendment + Cap-4-checker-extension + verification + rollback design
input_ruling: OD_EB1_decision = APPROVE_EVIDENCED_BY_AS_NEW_DURABLE_EDGE_TYPE
(apr_tier medium Đ32; approval_scope = design_APR_approval_only)
execution_authorized: false # sovereign ruling explicitly: execution NOT authorized
edge_in_real_vocab: false # vocab NOT amended
cap4_checker_changed: false # checker NOT modified
quality_gates: QG1–QG7 all addressed (see §6)
forbidden_respected: true
self_advance: PROHIBITED
verdict_requested_from: GPT / User (readiness for a future command-review)
The GPT-approved evidenced_by semantics are now expressed as: a precise logical vocab entry (reverse evidences, owner_law_code=NRM-LAW-44, family relation/Cap-4, default proposed); a six-family Cap-4 checker rule set (endpoint/OQC, authority, non-overlap, provenance, lifecycle, anti-drift); a structural verification list; a 12-row negative test matrix; and a failure-mode rollback/compensation plan. Nothing was executed. The evidenced_by ⟂ iu_entity_binding boundary is preserved end-to-end. All open decisions are flagged, not self-resolved.
2. What was read
| # | Document | Used for |
|---|---|---|
| 1 | Sovereign GPT ruling (reviews/…evidenced-by-APR-sovereign-gpt-ruling…) |
OD-EB1 APPROVE, strict boundaries, post-approval required work, binding usage rules, next-phase mandate |
| 2 | evidenced_by APR draft (3 files) |
Approved semantics, §3.3 mapping, why-not-existing, endpoint/binding boundary, WS-3 R6 examples |
| 3 | P44-4A design/05-relation-edge-conformance-design.md |
§3.3 steps 4–5, §3.1/§3.2 edge set, §3.4 reverse convention, §4.1–§4.5 E1–E8 + INV, §4.4 lifecycle, §2.1/§4.2 OQC, §2.3 namespace (P44-4-α) |
| 4 | WS-1/WS-2/WS-3 authority (folder) | authority_semantics roles + override, binding-vs-edge rule, WS-3 alt-B context, BP-WS2-2 |
3. What was designed
file_1_vocab_amend_design:
- logical edge_type vocab entry (NOT applied): evidenced_by / reverse evidences /
owner_law_code NRM-LAW-44 / relation Cap-4 / default proposed / provenance required
- reverse-type treatment (named-inverse, reverse-index served)
- namespace treatment: OPEN P44-4-α flagged STILL OPEN, not pre-empted
- lifecycle status proposal (P44-4A §4.4, unchanged enum)
- fit with 8 Core + 3 Candidate (1 Extension; no existing def changed; Cap-5 untouched)
file_2_cap4_checker_extension_design:
- rule families R-EP / R-AU / R-NO / R-PV / R-LC / R-AD (logical, no code)
- endpoint+OQC, no raw_entity; authority (normative source / evidence target, override-only)
- non-overlap rejections → references/implements/derived_from/governed_by/iu_entity_binding
- provenance + lifecycle validation; checker decision flow (logical)
file_3_verification_and_rollback_plan:
- SV-1..SV-10 structural checks; 12-row negative test matrix incl. positive control
- rollback/compensation for vocab-wrong / checker-wrong / bad-rows-later; ordering+guardrails
file_4_report: this document
nothing_executed: no vocab/config write, no code, no schema, no SQL, no Directus mutation,
no dry-run, no backfill, no data migration, no CUT/VERIFY, no git commit
4. Remaining open decisions (flagged — NOT self-resolved, QG7)
OD-VC1: OPEN P44-4-α — final edge_type namespace scheme (prefix obj./sem. vs owner_layer
tag vs both). Design tags owner_law_code=NRM-LAW-44 but does NOT decide the scheme.
OD-VC2: Physical shape of the vocab entry (dot_config row / framework record / JSON schema)
— deferred to command-review; not specified here.
OD-VC3: Cap-4 checker deployment mechanism (versioned rule set / reversible toggle) and
pre-state snapshot/backup strategy — command-review concern.
OD-VC4: BP-WS2-2 — governed_by/derived_from Candidate ratification (P44-4A v1.0, User).
evidenced_by does not depend on it, but R-NO3/R-NO4 reference those edges as
required alternatives; tracked, not resolved here.
OD-VC5: reverse-index passes P1/P2/P3 (P44-4A §6) remain deferred — affects how `evidences`
is served at scale; not closed here.
OD-VC6: authority_semantics (Fabric roles) vs P44 owner_law tagging — two distinct layers;
confirm treatment at command-review (R-AU uses Fabric roles, vocab uses owner_law).
OD-VC7: INV-E5/OQC-2 ">=2 status states with real rows" is a post-implementation production
audit obligation — owner/cadence to be set at command-review.
5. Readiness for a future command-review
ready_for_command_review_preparation: YES
rationale:
- APR semantics fully translated to a reviewable vocab entry + checker rule set
- verification (structural + 12-row negative matrix) and rollback/compensation defined
- evidenced_by ⟂ iu_entity_binding boundary preserved (QG3) incl. raw-entity routing (QG4)
- no execution; all open decisions flagged for sovereign decision (QG7)
NOT_ready_to_execute: command-review itself is a SEPARATE GPT/User-authorized phase;
OD-VC1..OD-VC7 should be addressed/acknowledged before or within that command-review
6. Quality gates
| Gate | Requirement | Status |
|---|---|---|
| QG1 | Do not repeat APR draft; only convert APR ruling → vocab/checker design | PASS — files consume the ruling; justification not re-argued |
| QG2 | No executable SQL/code | PASS — all rules/entries are logical specs / pseudo-rules; zero SQL/DDL/DML/code |
| QG3 | Preserve evidenced_by vs iu_entity_binding boundary | PASS — R-EP5/R-NO5, NT-1/NT-2, vocab §6 boundary |
| QG4 | Negative test matrix present | PASS — verification plan §3 (NT-1..NT-12) |
| QG5 | Rollback/compensation plan present | PASS — verification plan §4 (4.1–4.4) |
| QG6 | State execution still forbidden | PASS — gating banners + non-execution statements in all 4 files |
| QG7 | Flag all remaining open decisions, do not self-decide | PASS — §4 OD-VC1..OD-VC7; P44-4-α/BP-WS2-2 carried |
7. Recommended next step (advisory only — GPT/User decides)
options_presented (sovereign ruling §later_options + task menu):
A: schema_design_Q5 — registries + assembly_profile/topic/entity metadata physical
B: implementation command-review for vocab amend + Cap-4 checker (executes THIS package)
C: compliance_matrix logical proof using approved evidenced_by semantics
D: source/grammar ratification for the actual Constitution fixture (OD-G1/OD-S1)
recommendation: B (implementation command-review for vocab/checker) — advisory only
why_B:
- this package is purpose-built as the command-review input; B has the clearest,
most-bounded scope and unblocks A/C downstream
- A (schema Q5) is gated on BR-A1/BR-FA1/BR-FA3 still-open blockers
- C (compliance_matrix) is stronger once evidenced_by is actually live (post-B)
- D is independent (defer_post_pilot per WS-3 gap review) and not on the critical path
not_recommended_now: self-advancing into A/B/C/D — all are GPT/User-gated phases
8. Forbidden respected
vocab_amended: false
cap4_checker_code_changed: false
schema_migration: none
table_view_function_index_created: none
production_write: none
backfill: none
data_migration: none
cut: none
verify: none
dry_run: none
directus_mutation: none
vector_nosql_integration: none
git_commit: none
self_advance_to_command_execution_or_schema_or_compliance_matrix: none
agent_decision_on_open_items: none # OD-VC1..VC7 reserved for GPT/User
9. Git / repo access note
repo_vps_access: NONE in this session
basis: cwd /Users/nmhuyen is not a git repo; /opt/incomex/dot/iu-cutter absent;
no VPS helper present (verified in the immediately preceding phase, same session)
branch / HEAD / `git status --short -- iu-cutter`: UNAVAILABLE (no repo/VPS access)
write_operations: none
note: design-only KB authoring; no repo interaction required or attempted
10. Routing
status: EVIDENCED_BY_VOCAB_AMEND_AND_CAP4_EXTENSION_DESIGN_COMPLETE__DESIGN_ONLY
package: 4 files uploaded to knowledge/dev/laws/dieu44-trien-khai/v0.5-fabric-addendum-scope/
next_action: ROUTE TO GPT / USER for review + selection of next phase (A/B/C/D)
agent_next: STOP — execution NOT authorized; self-advance PROHIBITED
STOP. Design-only package complete. evidenced_by is approved at design-governance level only; vocab is NOT amended, the Cap-4 checker is NOT modified, nothing executed. Awaiting GPT/User review and next-phase selection.
End of report.