00 — Readme First

Macro: REGISTRIES_PIVOT_PARENT_GRAPH_COMMIT_AND_AGGREGATE_PIVOTS Date: 2026-06-03 Final status: ✅ PASS Execution mode: EXECUTION_MODE (read query_pg RO; write ssh contabo → docker exec postgres psql -U workflow_admin superuser RW) Live mutation: YES — the 13-edge parent_code graph is committed to production.

Headline

The Registries-Pivot dynamic layer graph is now LIVE and backend-resolved. pivot_definitions.parent_code was committed for 13 child pivots → the tree is hierarchical (24 roots / 6 parents / 13 children, 0 dangling / 0 cycle / 0 inactive-parent). v_registries_pivot_node_contract renders the hierarchy with zero view change. The layer graph no longer depends on naming convention — it is parent_code data.

What changed this macro

1. Committed the 13-edge parent graph (the single approved live mutation). Pre-image backed up (all 13 were NULL), rollback script saved.
2. Characterised the count-refresh side-effect exactly: only 1 stale cache value corrected (PIV-201 entities +3, stale→live); 0 meta_catalog changes, 0 L1 registry-total changes, 0 business-table touch.
3. Re-classified PIVOT_MISSING and proved (rehearsal) that PIV-311 (IU=219) and PIV-313 (sys-issues open=207,940) are READY_TO_ADD now — the engine already supports view/any-relation sources, so no engine extension is needed; the only hold is owner ratification (births land cleanly).

Document map

• 01 live state before commit
• 02 13-edge set + validation
• 03 live commit report (with side-effect proof)
• 04 tree + node contract after commit
• 05 aggregate pivots / PIVOT_MISSING reclassification
• 06 count integrity + accounting equation
• 07 UI handoff after commit
• 08 rollback + recovery package
• 09 done-enough decision
• 10 final summary
• 11 GPT MCP-readable checkpoint (in-package)
• standalone: checkpoint-rp-parent-graph-commit-2026-06-03.md

Compliance

No RP cleanup, no dot-pivot-update, no birth/governance/pre-birth alteration (aggregate-pivot births were rehearsed in BEGIN..ROLLBACK only), no fake pivots, no Nuxt count math, no hardcoded depth. The only count-content side-effect is the system's own designed statement-level refresh, documented exactly.