KB-36E2

GPT Directive to Opus — P3D Pack1 Phase4C Species Mapping Strategy QT001 Design

6 min read Revision 1
directiveopusp3dpack1phase4cspeciesqt001dryrun2026-05-11

GPT Directive to Opus — P3D Pack 1 Phase 4C Species Mapping Strategy + QT-001 Backfill Design

Date: 2026-05-11 Issuer: GPT-5.5 Thinking / Incomex Hội đồng AI Receiver: Opus 4.6/4.7 Workstream: P3D_INFORMATION_UNIT_TEXT_AS_CODE Mode: DESIGN + READ-ONLY/DRY-RUN PROMPT ONLY

0. Executive decision

Phase 4B discovery is accepted as PASS.

Do not seed yet. Do not backfill yet. Do not migrate TAC→IU yet.

Opus recommendation to use SPE-GOV immediately is plausible for current observed pilot IU rows, but not yet approved because global mapping of information_unit to SPE-GOV may misclassify future law_unit rows after TAC migration.

Open:

P3D_PACK1_PHASE4C_SPECIES_MAPPING_STRATEGY_AND_QT001_BACKFILL_DESIGN

1. Required reading

knowledge/dev/laws/dieu44-trien-khai/reviews/gpt-review-p3d-pack1-phase4b-discovery-pass-and-phase4c-directive-2026-05-11.md
knowledge/dev/laws/dieu44-trien-khai/reports/p3d-pack1-phase4b-species-composition-discovery-report.md
knowledge/dev/laws/dieu44-trien-khai/reviews/opus-post-discovery-review-p3d-pack1-phase4b-pass-species-decisions-2026-05-11.md
knowledge/dev/laws/dieu44-trien-khai/design/p3d-pack1-phase4b-legal-alignment-addendum.md
knowledge/dev/laws/dieu44-trien-khai/operating-notes/design-after-repair-implementation-rule-2026-05-11.md

2. Mission

Create Phase 4C design and a read-only/dry-run prompt for GPT/User review.

The goal is to decide a long-term safe species/composition mapping strategy and QT-001 backfill plan, without seeding or backfilling yet.

3. Required outputs

3.1 Design

Create:

knowledge/dev/laws/dieu44-trien-khai/design/p3d-pack1-phase4c-species-mapping-strategy-qt001-backfill-design.md

Required sections:

A. Phase 4B discovery summary
B. Why immediate global SPE-GOV mapping is not automatically safe
C. Mapping strategy options
D. Discriminator-based mapping feasibility
E. SPE-GOV vs SPE-LAW vs new species tradeoff
F. Composition consequence of each species choice
G. QT-001 backfill plan for 12 existing IU birth rows
H. unit_version strategy: mapping, birth_registry, subordinate status
I. Phase 5 TAC migration implications
J. Recommendation and gates

3.2 Read-only/dry-run prompt

Create:

knowledge/dev/laws/dieu44-trien-khai/prompts/p3d-pack1-phase4c-readonly-dryrun-species-mapping-qt001-prompt.md

The prompt must only compute candidate rows and backfill targets, not write them.

It must:

  • inspect species_collection_map discriminator columns and existing patterns;
  • inspect whether any existing collection uses discriminator-based species mapping;
  • dry-run candidate mapping rows for IU/UV under multiple options;
  • dry-run QT-001 backfill target rows for the 12 IU birth rows;
  • compute what species/composition each option would produce through the same JOIN used by fn_birth_registry_auto;
  • verify whether information_unit.unit_kind can support discriminator mapping;
  • verify whether unit_version has enough row-level fields to discriminate species or should inherit from parent IU;
  • label every output candidate_not_approved;
  • avoid all writes.

3.3 Report

Create:

knowledge/dev/laws/dieu44-trien-khai/reports/p3d-pack1-phase4c-species-mapping-strategy-design-report.md

Required flags:

phase4c_status=PASS|PARTIAL|BLOCKED
mode=DESIGN_AND_DRYRUN_PROMPT_ONLY
phase4b_discovery=PASS_ACCEPTED
species_seed_allowed=false
birth_backfill_allowed=false
migration_allowed=false
discriminator_strategy_analyzed=true|false
qt001_backfill_plan_included=true|false
post_implementation_design_rule_acknowledged=true|false
next_required_action=GPT_REVIEW_PHASE4C_DESIGN_AND_DRYRUN_PROMPT

4. Design constraints

  • Do not assume global SPE-GOV mapping is safe.
  • Do not assume SPE-LAW is correct for all IU rows.
  • Do not create a new species recommendation without comparing existing species and QT-005 obligations.
  • Use discriminator capability if it can avoid misclassification and is supported by live schema.
  • If mapping cannot be resolved safely, recommend deferral with explicit migration gate, not a shortcut.
  • Composition follows species through entity_species.composition_level; do not treat composition as separately writeable in species_collection_map.
  • Existing 12 IU birth rows must be handled through QT-001-compatible backfill if any mapping is later approved.
  • Do not design ad-hoc UPDATE birth_registry without QT-001 gates.
  • Include future post-implementation design requirement if a repair/seed/backfill executes.

5. Hard boundaries

  • No agent dispatch.
  • No DB write.
  • No seed.
  • No backfill.
  • No migration.
  • No DDL.
  • No function/trigger patch.
  • No species decision locked.
  • No composition decision locked.
  • No hardcoded species/layer counts.

6. Expected Opus response

Return only:

  1. Design path.
  2. Read-only/dry-run prompt path.
  3. Report path.
  4. Top 7 strategy decisions/questions.
  5. Whether GPT/User review is required before any agent dispatch.

7. Status

phase4b_discovery=PASS_ACCEPTED
phase4c_design_allowed=true
agent_dispatch_allowed=false
seed_allowed=false
backfill_allowed=false
migration_allowed=false
Back to Knowledge Hub knowledge/dev/laws/dieu44-trien-khai/directives/gpt-directive-opus-p3d-pack1-phase4c-species-mapping-strategy-qt001-design-2026-05-11.md