← All guides · Reviewed November 2026
Agentic Commerce Checklist for Online Stores
The short answer
A store is ready for agent commerce when an AI agent can reach it, read its catalog as structured data, check its policies, and walk to checkout without a forced login, a challenge page or a script-only control. This checklist is mapped item-for-item to Agent-Ready Store Audit's real checks across its five sections — work it top-down, then retest: the badge standard is 70+ overall with no critical fail. Checklist mapped to the audit engine's check set, November 2026. Score yours with the free store audit.
The checklist
| Section | Item | Audit check |
|---|---|---|
| Access | HTTPS is enforced on every page | https_enforced |
| Access | The homepage answers a plain HTTP read (no challenge wall) | homepage_reachable |
| Access | robots.txt exists and lets agent crawlers read the store | robots_present · robots_allows_agents |
| Access | sitemap.xml is reachable and lists product pages | sitemap_present · sitemap_lists_products |
| Access | Named agent bots get the same store as a browser (Pro) | bot_differential |
| Catalog | A machine-readable product feed answers | product_feed |
| Catalog | Product pages carry JSON-LD Product data | jsonld_present |
| Catalog | JSON-LD states price, currency and availability | jsonld_price · jsonld_currency · jsonld_availability |
| Catalog | Products carry identifiers (GTIN, MPN, SKU or brand) and descriptive titles | product_identifiers · product_title_descriptive |
| Catalog | Product images carry alt text; one currency is used across the page | image_alt_coverage · currency_consistency |
| Policies | Privacy and terms pages are reachable | legal_pages |
| Policies | Shipping policy, returns policy and a contact page are reachable | shipping_policy · returns_policy · contact_page |
| Purchase path | The cart page answers (simulation) | cart_reachable |
| Purchase path | The product page exposes an add-to-cart path (simulation) | add_to_cart_signal |
| Purchase path | A checkout entry point is reachable from cart or product (simulation) | checkout_entry |
| Agent readiness | llms.txt is present | llms_txt |
| Agent readiness | The UCP discovery file answers at /.well-known/ucp — and its claims match the live store (Pro) | ucp_wellknown · ucp_claim_vs_reality |
| Agent readiness | MCP readiness signals are present | mcp_signals |
Why each item matters — and where agents stall when one fails — is covered in Can AI Agents Buy From Your Store?
How do you work it?
- Access first. An agent that cannot read the store never sees the catalog. Fix HTTPS, robots and sitemap before anything else.
- Then the catalog. Structured product data is the agent's shelf label: price, currency, availability, identifiers.
- Then policies. A careful agent checks shipping, returns and contact before it commits its customer.
- Then the path. Cart, add-to-cart, checkout entry — reachable by plain reads, with no forced login or challenge in the way.
- Then readiness, then retest. llms.txt, UCP and MCP signals are the differentiators. Retest after fixes: the audit is deterministic, so a real fix flips the check, and the Pro validated fix pack exports the remaining items sorted by impact.
Run the free store audit · Ingenious AI
Frequently asked questions
What is on the agentic commerce checklist?
The checks an AI shopping agent's visit stands or falls on, in the audit's five sections. Access: HTTPS enforced, homepage answering a plain HTTP read, robots.txt present and letting agent crawlers read the store, sitemap present and listing product pages. Catalog: a machine-readable product feed, product pages carrying JSON-LD with price, currency, availability and identifiers. Policies: privacy, terms, shipping, returns and contact pages reachable. Purchase path: cart reachable, an add-to-cart path exposed, a checkout entry that resolves. Agent readiness: llms.txt, the UCP discovery file at /.well-known/ucp, and MCP readiness signals.
Why does robots.txt matter for agent sales?
Because it is the door notice. The audit fetches your robots.txt and tests whether named agent crawlers — GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended — may read the store, and separately whether those bots get the same store as a browser (the bot-differential check). A store that blocks or walls off agent crawlers has opted out of agent commerce before the first product is seen.
What product data does an agent need?
Structured, not visual: JSON-LD Product data on the product page stating a price, a currency and availability, product identifiers (GTIN, MPN, SKU or brand), a descriptive title, alt text on product images, and one currency used consistently across the page. An agent cannot buy from a price it cannot parse, however beautiful the page looks to a human.
How do I work the checklist?
Top-down, in section order — access before catalog, catalog before purchase path — then retest: the audit is deterministic, so the same store state gives the same result, and a fix shows up as a flipped check. The Pro validated fix pack turns the failing checks into exportable, checkable fix items sorted by impact.
Does passing the checklist guarantee agent sales?
No. It guarantees the agent can complete the visit: find the store, read the catalog, check the policies and reach checkout without stalling. Whether the agent's customer buys is still about the product and the price — the checklist removes the storefront as the reason the sale never had a chance.