Compare Shopify and NetSuite inventory by SKU

**Objective:** Compare Shopify inventory with NetSuite quantity available for {sku} (a single SKU, or leave blank for up to {max-records} stock rows, default 50), optionally limited to the NetSuite location {location-name}.

**Step 1: Read stock from Shopify**

Call `shopify-get-inventory-level` (Shopify server) with `sku` set to {sku} when given and `max-records` set to {max-records}. Keep `sku`, `inventory-item-id`, `location-id` and `available`. Call `shopify-get-location` (Shopify server) to turn each `location-id` into a location `name`. Sum `available` per SKU across locations unless {location-name} is given.

**Step 2: Read stock from NetSuite**

For each SKU call `netsuite-get-itemquantity` (NetSuite server) with `item-number` set to the SKU and, if {location-name} is given, `location-name` set to it. Keep `quantity-on-hand` and `quantity-available`. Sum per SKU across bins and locations unless a location was given. Compare the Shopify `available` with the NetSuite `quantity-available`.

**Step 3: Report**

A table of SKU, Shopify available, NetSuite quantity available, NetSuite quantity on hand and the difference, with the largest absolute differences first. Group SKUs as matching, different, in Shopify only and in NetSuite only. End with counts per group.

**Rules & Guidelines:**

* Tools whose names start with netsuite- are on the NetSuite server; tools whose names start with shopify- are on the Shopify server. Both servers must be connected.
* This is a read-only job. Do not call any tool whose name starts with netsuite-add-, netsuite-update-, shopify-add- or shopify-update-, and never create, change or delete a record on either side.
* Report only what the tools return. Do not estimate or invent figures; totals are sums of rows you actually retrieved, and say so.
* If a read returns the maximum number of rows, say the result may be truncated and suggest narrowing the filters.
* Dates are YYYY-MM-DD. Where an input is blank, use the default stated in the Objective.

**Validation:**

* If the first read returns nothing, say so and stop rather than guessing.
* If a lookup on the other system returns more than one record, list them and mark the row ambiguous instead of choosing one.
* If a lookup returns nothing, report the record as present in one system and absent from the other, and continue with the next row.
* If the two systems disagree on a value, show both values side by side and do not decide which is right.