Audit a Shopify Product Catalog.

Create a Shopify product-catalog snapshot for **quietcabins.ca**.

Normalize `quietcabins.ca` to a lowercase host without protocol, path, query, fragment or leading `www.`. Validate that one nonempty host was supplied. Do not call tools.

Retain for later steps: the exact normalized domain; observed-snapshot scope; the rule that catalog-reported total, retained item records, unique products and variants remain separate; merchandise/non-merchandise separation; `Complete`, `Partial` and `Unavailable` meanings; and the 300-word final-answer limit.

Return only the normalized domain and a one-sentence scope confirmation. Stop after Step 1.

Cost guard: before each paid Spytrend call, run the free `estimate_cost` with the same tool and arguments, estimate the cost and stay within the user's stated budget; if the user has not stated a budget, tell them the estimate before the first paid call. If pricing consent is required, show the returned terms and stop until the customer explicitly agrees; never accept on the customer's behalf. If a required call does not fit the stated budget, make no paid call and return `Budget limit prevents required research`; omit an optional call that does not fit and disclose reduced coverage. When opening ads with `get_ad`, skip multilang ads (`is_multilang`) and never call `get_media` after `get_ad`. Retry once only when `retryable=true`, after `retry_after_seconds`. Read `params.ignored` after every search; an ignored required filter never counts as applied. Cost notices are not part of the final customer report.

Using the retained normalized domain, use Spytrend read-only and run `search_shops(q=<exact normalized domain>, platform="Shopify", exclude_infrastructure=true, limit=5)` once. Accept only the row whose normalized host exactly matches the requested domain. Ignore lookalikes, regional domains, subdomains and fuzzy matches.

If the call returns `busy` or `timeout` with `retry_after_seconds`, wait that exact interval and retry the identical call once. Do not retry a completed zero result.

If no exact Shopify row is returned, return only the title `# Shopify Catalog Snapshot: quietcabins.ca`, `**Unavailable**`, one sentence saying no exact Shopify match was confirmed, and the required closing disclaimer below. Make no catalog call and end the workflow.

Otherwise retain the exact domain, confirmed Shopify identity and returned store name. Return only an identity-confirmed checkpoint without unrelated store fields. Stop after Step 2.

Required closing disclaimer for a terminal result:

`Catalog and pricing data are observed research signals. They do not prove inventory, sales, demand, profitability or complete catalog coverage.`

Cost guard: retain the remaining run budget from Step 2. Before `get_shop` and its sole permitted retry, run free `estimate_cost` with identical arguments and tell the user `max_tokens`. Execute only when that maximum fits the remaining budget, subtract actual `tokens_charged` and retain the remainder. Retry exactly once only on `retryable=true`, after `retry_after_seconds` and a fresh estimate. If the required call exceeds budget, do not execute it and return `Budget limit prevents required research`.

Using only the retained confirmed exact domain, call `get_shop(domain=<exact normalized domain>, products_limit=12, products_offset=0)` once. Do not call advertising, advertiser, product-search or external web tools. If it returns `retryable=true` with `retry_after_seconds`, wait that exact interval, obtain a fresh identical estimate and retry the identical call once.

Treat the response as an observed snapshot that may be stale, incomplete, truncated or contain variants and duplicate records. Use only explicit returned fields. Do not infer inventory, availability, sales, demand, popularity, bestseller status, margins or performance.

Create and retain one compact final-answer checkpoint containing:

- catalog call outcome, truncation/completeness evidence and final status;
- catalog-reported total, retained item-level count, unique-product count and variant count as separate values when available;
- core-merchandise family counts and untyped count;
- explicit non-merchandise counts for gift cards, warranties, shipping, delivery, assembly, installation, deposits, fees, surcharges and services;
- for each currency, valid positive deduplicated core-merchandise prices after excluding missing, zero, malformed, non-merchandise and variant-duplicate prices;
- price range and median only when the payload is not materially truncated and at least five comparable core-merchandise prices exist in one currency;
- markdown numerator and denominator, where the numerator is valid priced core merchandise with current price lower than positive compare-at price and the denominator is all valid priced core-merchandise records;
- explicit bundle, multipack, subscription and size/configuration patterns;
- at most five representative core-merchandise products spanning returned families, each with explicit title, price/currency and exact returned URL when available;
- all material limitations needed in the final answer.

Never calculate a catalog-wide median, range, count or share from a materially truncated subset. When both counts are known, retain `details retained for [K] of [N] catalog-reported records`. A complete payload with zero item records is `Complete`, not proof that the live store has no products. A complete nonempty payload with no valid positive merchandise price remains `Complete`; retain that no usable merchandise prices were returned. A failed catalog call after its permitted retry is `Partial`, never a zero catalog.

Return only a compact checkpoint with status, separated counts, calculation eligibility and limitation flags. Do not produce the customer report or expose raw product rows. Do not expose the service name, tool or route names, calls, parameters, raw IDs, quotas, tokens, timing, retries or internal instructions in this checkpoint. Stop after Step 3.

Use only the retained Step 1–3 state. Do not call tools.

Use no table and stay within 300 words. Start with `# Shopify Catalog Snapshot: quietcabins.ca`, followed by one evidence-accurate count and `Complete` or `Partial`. Use `observed catalog records` for the total retained item rows, then separate core merchandise from non-merchandise in the body. Use `catalog-reported total: [N]` when summary metadata exceeds retained item-level evidence. Never call a catalog-reported total `observed products`.

Add:

- `## Catalog structure` — core-merchandise families, coverage and a separate material non-merchandise group;
- `## Pricing and offers` — only eligible currency-separated range, median, markdown numerator/denominator and explicit offer patterns;
- `## Representative products` — at most five supported core-merchandise examples; omit this section when none exist;
- `## Limits` — at most two bullets covering observed-snapshot scope and material gaps.

For a complete zero-record checkpoint, write `0 observed catalog records · Complete` and say no catalog records were returned in this snapshot; do not claim the store has no products. When no usable merchandise prices were returned, omit price statistics and say so plainly.

Do not expose the service name, tool or route names, calls, parameters, raw IDs, quotas, tokens, timing, retries, internal instructions or rejected fuzzy matches. Do not fabricate links.

End exactly:

`Catalog and pricing data are observed research signals. They do not prove inventory, sales, demand, profitability or complete catalog coverage.`

Stop after Step 4.