How we check
How we check before telling you to act
We answer merchant questions with primary sources. We leave out announcements that do not change what a merchant should do.
What enters
A change enters only when it alters a merchant deadline, permission, supported path, cost, required action, or trustworthy interpretation of store data.
What stays out
Vendor launch language, unsourced predictions, duplicate coverage, developer changes with no merchant consequence, and metrics without a merchant question stay out. A release can be large and still require no guidance.
How a source passes
- Open the primary changelog, specification, policy, help page, or filing.
- Record the date it was checked.
- State exactly when the answer applies and when no action is needed.
- Write the Official Path, Steps, Done condition, and Stop Point when work is required.
- Keep the source link on the published page so the reader can verify it.
How publication is controlled
Scheduled collectors can write only to content-candidates/, which sits outside the site’s content collections. Their credentials and write permissions must exclude both published directories. A human publishes an item by moving it into the English and Traditional Chinese collections in a reviewed change. Candidate fields such as draft or reviewedDate have no publishing authority and are rejected in published files.
How published content stays current
Every published item records a Source Check. Platform Changes must be checked again within 30 calendar days. Decisions and Steps must be checked within 90 calendar days. A stale or future-dated item fails the next site build. English and Traditional Chinese entries must have matching identifiers, dates, categories, Steps, and source evidence.
How lists are ordered
The complete guidance list, answers, and steps use the original publication date. Platform changes use the date when the change took effect. This keeps a later write-up from making an older change appear new.
Automation boundary
Published content accepts Markdown only. Raw HTML is rejected. Every external link must match an exact HTTPS URL in the validated source register. Every answer states when it applies, when no action is needed, why it matters, what to do, how to know it is done, and what the result does not prove. Each answer links to evidence already declared in the article’s source list.
Registered sources
Other sources may point to a change, but they do not replace its primary source. The registered sources and planned review cadence are:
- Shopify: Merchant changelog. Planned review cadence: daily. Merchant-facing launches, deadlines, defaults, and admin workflow changes.
- Shopify: Developer changelog. Planned review cadence: weekly. Commerce protocols and developer changes that alter a merchant-supported route.
- Shopify: Agentic storefronts help. Planned review cadence: weekly. AI-channel eligibility, checkout behavior, product support, privacy, and opt-out controls.
- Shopify: AI-powered tools help. Planned review cadence: weekly. Sidekick, official assistant connections, permissions, and merchant review boundaries.
- Shopify: Shopify Flow help. Planned review cadence: weekly. Workflow setup, testing, supported actions, and automation safety.
- OpenAI: Shopping with ChatGPT Search. Planned review cadence: weekly. Current shopping-result, merchant-ranking, price, review, and checkout behavior.
- OpenAI: Commerce policies. Planned review cadence: weekly. Prohibited products and practices that alter merchant eligibility.
- Google: Merchant Center announcements. Planned review cadence: daily. Merchant Center deadlines, reporting changes, new attributes, and AI commerce availability.
- Microsoft: Microsoft Advertising agentic commerce. Planned review cadence: weekly. Copilot Checkout eligibility, onboarding, fees, and merchant controls.
- Perplexity: Instant Buy help. Planned review cadence: weekly. Merchant responsibilities, eligibility, product data, and checkout mechanics.
- OpenAI: Powering product discovery in ChatGPT. Planned review cadence: weekly. How product data reaches ChatGPT shopping experiences.
- Shopify: ChatGPT agentic storefront. Planned review cadence: weekly. Shopify eligibility, catalog syndication, and checkout behavior in ChatGPT.
- Shopify: Connecting AI tools to a store. Planned review cadence: weekly. Official assistant connections, permissions, and access controls.
- Google: AI performance insights. Planned review cadence: weekly. Merchant Center AI visibility metrics, eligibility, and interpretation limits.
- Google: AI performance insights announcement. Planned review cadence: weekly. Dated availability announcement for the Merchant Center pilot.
- Ahrefs: llms.txt server-log study. Planned review cadence: monthly. Original empirical evidence about requests to published llms.txt files.
- Shopify: llms.txt, llms-full.txt, and agents.md customization. Planned review cadence: weekly. Shopify storefront support for machine-readable discovery files.
- OpenAI: Commerce product feed specification. Planned review cadence: weekly. Current production and draft product feed fields and submission requirements.
- Shopify: Agentic Storefronts in Shopify admin. Planned review cadence: weekly. Admin reporting and recommendations for agentic channels.
- Shopify: Spring 2026 merchant edition. Planned review cadence: monthly. Merchant-facing release details for agentic commerce reporting.
- Shopify: Agentic commerce audit. Planned review cadence: monthly. Shopify's free product-page audit and its stated interpretation limits.
- Google: Analytics default channel definitions. Planned review cadence: monthly. Google Analytics classification of traffic from AI assistants.
- Shopify: Perplexity shopping integration. Planned review cadence: monthly. Shopify Catalog syndication and checkout mechanics for Perplexity.
- Anthropic: Anthropic crawler controls. Planned review cadence: monthly. Crawler identities, purposes, and robots.txt behavior.
- Shopify: Storefront MCP cart tools deprecation. Planned review cadence: weekly. UCP migration deadline for integrations using Storefront MCP cart tools.
- Shopify: Global Catalog MCP. Planned review cadence: weekly. Documented agent access to Shopify's cross-merchant catalog.
- Shopify: WebMCP for Liquid and Hydrogen. Planned review cadence: weekly. Built-in WebMCP availability and production status by storefront stack.
- Shopify: Test workflows in Shopify Flow. Planned review cadence: monthly. Workflow test behavior, supported data, and action limitations.
- Shopify: Sidekick-generated Flow test cases. Planned review cadence: monthly. Sidekick support for generating Shopify Flow test cases.
- Shopify: Products in agentic storefronts. Planned review cadence: weekly. Product eligibility and discovery controls for Shopify agentic storefronts.
- OpenAI: Stable commerce product feed reference. Planned review cadence: weekly. Production-supported fields and formats for OpenAI product feeds.
- OpenAI: Draft commerce product feed reference. Planned review cadence: weekly. Planning-only fields under consideration for a future OpenAI product feed.
- Shopify: Editing product variants. Planned review cadence: monthly. Standard Shopify variant fields and editing behavior.
- Shopify: Cart MCP. Planned review cadence: weekly. Current UCP-conforming cart tools and migration target.