ALLSoftThemes & apps
Open navigation

Configure / 04

Machine-readable storefront context, without false promises.

Citera can expose native request paths and structured data that help compatible systems understand public storefront content. These surfaces do not guarantee indexing, citation, ranking, or sales.

ReferenceMerchant facing
Applies toVersion 1.2.0
StatusPreview

Agentic commerce and GEO

Use one accurate commerce source across both discovery paths.

Shopify Catalog supplies structured product data to eligible agentic channels. Public product pages, policies, structured data, and discovery files support open-web understanding and GEO. They work together, but one does not replace the other.

Shopify Catalog

Shares product titles, descriptions, options, images, price, availability, and other catalog attributes with active channels.

Public storefront

Gives people and crawlers complete product facts, policies, comparisons, useful copy, and accessible purchase paths.

Knowledge Base

Gives supported AI shopping agents concise, maintained answers about shipping, returns, accounts, and store-specific questions.

Merchant setup

Prepare the Agentic channel before testing an assistant.

  1. In Shopify admin, open Sales channels > Agentic. Review the supplemental terms and the channels available to the store.
  2. Confirm whether Allow Shopify to manage for me is appropriate. If channels are managed manually, verify Catalog access for each intended destination.
  3. Complete Terms of service, Privacy policy, and Return and refund policy in Settings > Policies.
  4. In the Agentic Sources area, preview how representative products are sent to Shopify Catalog.
  5. Use Catalog Mapping only when the preferred title, description, category, option, or grouping source lives in metafields, metaobjects, tags, or another custom structure.
  6. Install Shopify Knowledge Base, review generated facts, answer top unanswered questions, and test the matched sources.

Catalog quality

Give the assistant facts it can use to make a recommendation.

Identity and intent

Use a clear product title, accurate category, concise summary, specific use cases, and real differentiators.

Decision facts

Publish dimensions, materials, compatibility, sizing, care, included items, warranty, and relevant legal disclosures.

Commerce truth

Keep options, variants, price, availability, market access, shipping, returns, and the product URL current.

Useful media

Use clear product imagery, honest crops, descriptive alternative text, and media that explains scale or variation.

AI assistant catalog test

Test customer intent, not only exact product names.

First use the Catalog preview in Shopify admin and the Test your knowledge base tool. Then run a small external assistant test where the channel is available. Use non-branded questions that describe a need, constraint, comparison, or policy question.

01

Home and interiors

Copy, adapt, and record the result for the target market.

  • Show me a neutral throw for a warm minimalist living room.
  • Which dining chair is suitable for daily use in a small apartment?
  • Compare two side tables by dimensions, material, finish, and price.
02

Fashion and beauty

Copy, adapt, and record the result for the target market.

  • Find a breathable option for warm weather in my size.
  • Which product is suitable for sensitive skin and how should it be used?
  • Compare these options by material, care, available colors, and returns.
03

Tools and technical catalogs

Copy, adapt, and record the result for the target market.

  • Which product fits this use case and what specifications prove it?
  • Compare the compatible models, included accessories, and warranty.
  • Show only available options that can ship to my location.

Record five facts for every test

  1. The exact prompt, date, market, language, and currency.
  2. Which products appeared and whether they matched the intent.
  3. Whether title, image, price, availability, and options were correct.
  4. Whether store policies and product claims matched their sources.
  5. Whether the product link opened the correct variant and checkout path.

Fix the source field that is wrong, then allow time for Catalog processing before retesting. Do not add repetitive keywords or hidden text to force a result.

Native request paths

Three explicit discovery surfaces.

/llms.txt

Concise storefront guide

A compact overview of public store identity, important routes, and catalog context.

/llms-full.txt

Expanded public context

A broader machine-readable view for supported public storefront content.

/agents.md

Agent guidance

Public usage notes and boundaries for compatible automated systems.

Where to edit

Shopify's builder exposes status, not long-form prose.

Theme settings show whether native discovery templates are included. Advanced text is maintained in the corresponding Liquid templates under Edit code:

  • templates/llms.txt.liquid
  • templates/llms-full.txt.liquid
  • templates/agents.md.liquid

This keeps the visual builder concise and prevents large technical text fields from crowding ordinary merchant settings.

Structured data

Enable only one authoritative schema source.

Citera can emit organization, website, product, breadcrumb, FAQ, article, blog, and collection-oriented structured data where the underlying content supports it. Disable overlapping theme output when a dedicated schema app is the chosen source.

  1. Keep storefront facts accurate and visible to customers.
  2. Enable the theme's structured data or a dedicated app, not conflicting duplicates.
  3. Validate rendered pages after product, policy, domain, or app changes.
  4. Never add hidden claims solely for crawlers or automated systems.

Official Shopify references

Recheck channel rules before each rollout.

Next guideReview release notes