Add multi-location caveat (Kimi second-model confirm) to remediation

31761c36f61b58d45aa1faa343e1a2b3ac8d37da · 2026-08-30 09:41:01 -0700 · steve@designerwallcoverings.com

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Files touched

M REMEDIATION.md

Diff

commit 31761c36f61b58d45aa1faa343e1a2b3ac8d37da
Author: steve@designerwallcoverings.com <steve@designerwallcoverings.com>
Date:   Sun Aug 30 09:41:01 2026 -0700

    Add multi-location caveat (Kimi second-model confirm) to remediation
    
    Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---
 REMEDIATION.md | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/REMEDIATION.md b/REMEDIATION.md
index ce0b328..d624b9d 100644
--- a/REMEDIATION.md
+++ b/REMEDIATION.md
@@ -18,6 +18,8 @@ This is the exact inverse of the importer's `setInventory2026()`; it lands each
 
 **Acceptance test:** `node verify.mjs` returns **exit 0 / RESULT: CLEAN** (0 $0-orderable across all scoped vendors), and a canary re-run reports **PASS**.
 
+> **Multi-location caveat (second-model confirmed, Kimi k2.5):** `availableForSale` aggregates stock across **all** online-fulfillment locations. `setInventory2026()` only ever wrote the Ventura Blvd location, so single-location zeroing is expected to suffice — but the apply script MUST verify no *other* location holds positive stock for the variant (and zero those too) before declaring clean. `verify.mjs` inherently catches this: it reads the aggregated `availableForSale`, so a residual location keeps the variant DIRTY.
+
 **Gate:** canonical, customer-facing Shopify write across 1,743 products → **HARD GATE** (>500 items + customer-facing). Draft only; needs Steve's explicit APPROVE.
 
 ## Fix B — Prevention (stop recurrence at the source) — RECOMMENDED, still gated for deploy

← 8b81bcc TK-10965: zero-price-orderable root-cause analysis + verify · back to Tk 10965 Zero Price Analysis · Add apply-fix.mjs (restore-map-first, canary-batched, rollba 7eb582b →