Start with a real store task. These 65 original prompts help you define the work, supply the right inputs, and review what an AI tool produces.
Published November 1, 2026 · AdsX Team · Free to use, no email required
These are illustrative briefs, not tested builds or guaranteed capabilities. This page does not connect to your store or run prompts. Use only information you are authorized to share; never paste credentials or unnecessary customer data.
Choose the right tool before copying
Sidekick-generated apps run inside Shopify admin. Theme prompts need a coding tool and an unpublished theme copy; a general assistant can help plan but does not automatically have store access. Review current Sidekick eligibility and limits and the Shopify AI Toolkit documentation. The tool label is a suggested environment, not a promise that every requested feature is supported.
These guides pair copyable prompts with current plan prices and worked cost examples. Subscription costs below use 12 monthly payments, before tax; they are potential avoided charges, not guaranteed net savings. Check free plans first and subtract setup, maintenance and any extra tools.
Kiwi: basic size chart display — $7.99/month Premium, or $95.88 over 12 payments. Keep specialist sizing functions when needed.
Bring: App names, bills, used features and connected systems
Map each supplied Shopify app to the functions we actually use. Separate storefront display, admin work, automation and external integrations. Identify overlapping features and missing dependency information. Recommend what to investigate before keeping, replacing or removing each app. Do not recommend uninstalling anything yet.
Expected output: A feature and dependency table
Review: Compare the table with app settings and ask the people using each workflow.
Illustrative prompt — not execution-tested
Planning · General assistant
Compare build versus buy
Bring: Subscription cost, build estimate, maintenance hours and hourly cost
Compare keeping this app with building the limited replacement described here. Calculate twelve-month cash costs and show labor separately. Include hosting, maintenance, support and update assumptions. Identify which unknown could reverse the decision. Do not invent a savings figure.
Expected output: An assumption-based cost comparison
Review: Recalculate the arithmetic and distinguish hypothetical inputs from actual quotes.
Illustrative prompt — not execution-tested
Planning · General assistant
Write a narrow app brief
Bring: One merchant problem and three representative examples
Turn this merchant problem into a one-page app specification. Define its user, data inputs, visible output, read or write actions, exclusions, failure states and acceptance cases. Recommend the appropriate Shopify surface. Ask about missing requirements before proposing implementation.
Expected output: A reviewable specification
Review: Make sure the specification excludes unrelated features and names the platform surface.
Illustrative prompt — not execution-tested
Admin apps · Sidekick Generate app
Build a catalog QA admin app
Bring: Product statuses to include and the definition of missing data
Generate a read-only Shopify admin app listing products without a meaningful description or product image. Show title, status and missing fields, with a way to review the product. Handle pagination, loading, failures and an empty result. Do not modify products or send data externally. Explain unsupported requirements and permissions before installation.
Expected output: An app draft and permission explanation
Review: Compare known complete and incomplete records, including one beyond the first results page.
Illustrative prompt — not execution-tested
Admin apps · Sidekick Generate app
Draft a team task tracker
Bring: Task fields, team workflow and storage requirements
Plan a simple Shopify admin task tracker for our internal team. We need task title, owner, due date and status. First explain whether the required storage and team behavior are supported in this environment. Flag unsupported features. Do not use customer records as task storage or request unrelated permissions. Build only after the plan is reviewed.
Expected output: A feasibility plan, then a scoped app draft
Review: Verify persistence, staff access and required fields; do not assume a visible list is durable storage.
Illustrative prompt — not execution-tested
Admin apps · Sidekick Generate app
Plan a reorder review dashboard
Bring: Date range, inventory locations and supplier lead times
Plan a read-only reorder review dashboard using supported store data. Explain the sales period, stock measure and proposed calculation. Identify missing supplier lead times and stock adjustments. Show assumptions beside results. Do not create purchase orders or change inventory. Flag unsupported data before building.
Expected output: A dashboard plan with data gaps
Review: Check calculations on known products and handle returns, stockouts and location differences.
Illustrative prompt — not execution-tested
Development · Coding agent
Find missing app requirements
Bring: App specification and relevant repository files
Review this specification against the current Shopify documentation and this repository. List missing API fields, permissions, persistence requirements and failure behavior. Separate supported implementation paths from assumptions. Do not modify files or run store mutations during this review.
Expected output: A gap list with source references
Review: Open the cited documentation and resolve blocking questions before implementation.
Illustrative prompt — not execution-tested
Development · Coding agent
Write acceptance cases
Bring: Feature brief and sample records
Create acceptance cases for this feature: normal use, empty data, invalid inputs, permission failure and partial network failure. State expected output and evidence for each case. Use supplied fixtures rather than private customer data. Do not claim tests have run.
Expected output: A test matrix
Review: Ensure cases test the desired behavior rather than merely repeat implementation details.
Illustrative prompt — not execution-tested
Development · Coding agent
Review requested permissions
Bring: Manifest, routes and feature list
Compare the requested Shopify permissions with the code paths and feature requirements provided. Explain each scope and identify unused or excessive access. Distinguish read from write access. Do not remove a scope automatically; list the impact and a verification step.
Expected output: A permission review
Review: Confirm every required action remains possible and no secret is included in the report.
Illustrative prompt — not execution-tested
Development · Coding agent
Prepare a rollback plan
Bring: Deployment method, changed files and data changes
Write a rollback plan for this change. Separate code rollback from data recovery. Identify backups, compatibility concerns, the person who can stop rollout and the evidence needed to resume. Do not assume reverting code reverses stored data changes.
Expected output: A release and recovery checklist
Review: Verify that recovery steps exist in the actual environment before deployment.
Illustrative prompt — not execution-tested
Theme · Coding agent with theme access
Draft a size-chart section
Bring: Unpublished theme, size table, product mapping and units
In the unpublished theme copy, propose a size-chart section using the supplied measurements and units. Inspect existing theme patterns first. Include accessible headings and mobile behavior. Keep chart assignment explicit per product. Do not publish the theme or invent measurements. Explain how the merchant edits the data.
Expected output: A theme change proposal or draft
Review: Review different products, long labels, keyboard access and narrow screens before publishing.
Illustrative prompt — not execution-tested
Theme · Coding agent with theme access
Draft an announcement bar
Bring: Unpublished theme, approved text and destination
Inspect this theme for an existing announcement setting before adding code. If a new section is needed, build a simple text-and-link announcement bar in the unpublished copy using existing conventions. Include editor settings and readable mobile styling. Do not add tracking, timers or fabricated urgency. Do not publish.
Expected output: A native-setting recommendation or small section draft
Review: Check the destination, contrast, keyboard access and behavior when the text is empty.
Illustrative prompt — not execution-tested
Content · Sidekick or general assistant
Write product copy from facts
Bring: Approved specifications and brand voice
Draft a concise product description using only these supplied facts. Separate included items from optional accessories. Preserve dimensions, materials and compatibility exactly. List unanswered buying questions separately. Do not add certifications, performance claims or delivery promises. Return a draft without publishing.
Expected output: A factual product-description draft
Review: Compare every claim, quantity and unit with the supplied specifications.
Illustrative prompt — not execution-tested
Content · Sidekick or general assistant
Draft product FAQs
Bring: Real questions, approved product facts and policies
Turn these real customer questions into a short product FAQ. Answer only from the supplied facts and current policies. Mark unanswerable questions for review. Avoid repeating the description and do not invent guarantees. Return draft text only.
Expected output: A question-and-answer draft
Review: Check policy dates and ensure each answer resolves the actual question.
Illustrative prompt — not execution-tested
Content · General assistant
Build a product comparison
Bring: Product fact sheets and comparison criteria
Compare these products using the supplied facts and criteria. Use the same units, mark unavailable information and keep opinion separate from fact. Do not pick a winner without a stated customer need. Return a table and the questions a buyer should resolve.
Expected output: A comparison table
Review: Verify source facts and avoid treating missing information as a negative product feature.
Illustrative prompt — not execution-tested
Content · Vision-capable assistant
Describe product images
Bring: Product images and their page context
Draft concise alt text for each supplied image based on what is visible and its purpose on the page. Distinguish product color or angle where useful. Do not infer material, certification or included accessories from appearance alone. Flag decorative images for human review.
Review: Inspect each image and avoid unnecessary repetition or keyword stuffing.
Illustrative prompt — not execution-tested
Planning · Sidekick or general assistant
Plan collection merchandising
Bring: Collection goal, product data and constraints
Propose a collection merchandising brief for this goal using the supplied products. Explain grouping, ordering and exclusions. Identify stock or seasonality assumptions and distinguish recommendations from verified facts. Do not move or publish products.
Expected output: A merchandising proposal
Review: Review available stock, customer intent and business constraints before applying changes.
Illustrative prompt — not execution-tested
Theme · Browser-capable coding agent
Review mobile shopping
Bring: Preview URL and target device widths
Review this preview as a mobile shopper. Inspect product options, price visibility, purchase controls, images and navigation at the supplied widths. Report observable issues with evidence and reproduction steps. Do not place orders or change the site. Distinguish screenshots from interactions actually tested.
Expected output: An evidence-based issue list
Review: Reproduce findings and prioritize blockers over cosmetic preferences.
Illustrative prompt — not execution-tested
Planning · General assistant
Check a discount brief
Bring: Eligibility, combinations, dates, currency and examples
Review this proposed discount specification for ambiguity. Work through the supplied carts, eligible products, combination rules and date boundaries. List expected outcomes and missing facts. Do not create a discount or assume the platform supports an unstated rule.
Expected output: A discount acceptance matrix
Review: Validate supported rules in Shopify and confirm totals before activation.
Illustrative prompt — not execution-tested
Planning · Sidekick or general assistant
Explain a store report
Bring: Authorized report, date range and metric definitions
Analyze this report using its stated date range and metric definitions. Separate traffic, orders, revenue, refunds and contribution where available. Identify missing denominators and reporting delays. Show calculations, distinguish correlation from causation, and suggest one useful next investigation.
Expected output: A bounded report analysis
Review: Check totals, time zones, comparison periods and whether revenue is gross or net.
Illustrative prompt — not execution-tested
App replacements · General assistant
Format approved size chart measurements
Bring: Approved measurements, units and definitions
Format these approved garment measurements into an accessible HTML table for a Shopify page: [measurements]. Preserve every number, unit, size label and measurement definition exactly. Include a caption explaining whether these are body or garment measurements. Flag missing or contradictory values. Do not estimate sizes, add fit recommendations, or convert units unless I supply an approved conversion rule. Return the table and a list of unresolved questions.
Expected output: An accessible chart and unresolved questions
Review: Compare every number and unit with the approved source.
Bring: Product handles and approved chart page handles
Given this product-handle list and these approved chart page handles, propose a product-to-chart mapping. Explain each rule. Flag ambiguous products and products that should have no chart. Keep adult, child and different garment cuts separate. Output a review table only; do not update Shopify or infer measurements.
Expected output: A proposed mapping table
Review: Check every ambiguous product and a sample from each rule.
Bring: Unpublished theme and page-reference metafield definition
Inspect this unpublished Shopify theme and its product template. We need a size-chart link sourced from a product page-reference metafield. Prefer the existing compatible pop-up block. If custom code is necessary, propose the smallest change, hide the link when no page is assigned, preserve keyboard focus behavior, and keep the table usable on narrow screens. Show the files, preview steps and rollback instructions. Do not publish, alter product data, or call an external service.
Expected output: A minimal theme proposal
Review: Check blank fields, product mapping, mobile tables and keyboard focus.
Bring: Approved message, existing URL and unpublished theme
In this unpublished theme, check whether the built-in announcement bar supports a text message and link. If it does, prepare the message “[approved message]” linking to [existing policy URL]. Preserve the approved qualification and do not change shipping rules, discounts or checkout settings. Show me the proposed change and preview. Do not publish the theme. If the feature is unavailable, explain the limitation instead of claiming it is configured.
Expected output: A supported native configuration or a limitation report
Review: Check the actual edited theme, message and destination before saving.
Inspect this Shopify theme. Propose an editable announcement section with one text field and one optional link, using existing theme styles. Keep it static, accessible by keyboard, and readable at 320px width and 200% zoom. Prevent it from covering navigation or content. Do not add tracking, geolocation, countdowns, external scripts or automatic discount logic. Show the minimal file changes, preview steps and rollback. Work only on the unpublished copy.
Expected output: An editable text-and-link section
Review: Test navigation, keyboard access, narrow screens and zoom.
Bring: Approved shipping or promotion rules and banner text
Review this banner text against these approved shipping and promotion rules: [rules]. Identify missing regions, thresholds, exclusions or end dates. Suggest the shortest accurate message and an existing destination link. Do not invent urgency, extend an offer or change its terms. List what must be verified at checkout.
Expected output: Accurate message options and checkout checks
Review: Confirm thresholds, regions and dates against actual settings.
Review these product tabs and approved content sources: [inventory]. Classify each as shared policy, product-specific text, conditional content, or another app’s output. Propose a content map for a Shopify theme. Flag missing source text and conflicting rules. Do not rewrite policy terms, infer care instructions, copy reviews, or change product records. Return a mapping table for review.
Expected output: A content ownership and destination map
Review: Check policy ownership, product differences and review integrations.
Bring: Unpublished theme and proposed field mapping
Inspect this unpublished Shopify theme and the proposed content map. Determine whether its existing collapsible rows can connect to compatible product metafields. Identify exact settings and field types from the theme and current documentation. Do not assume every block accepts every metafield type. Explain how blank values and shared content will behave. Prefer configuration over code and do not publish or change shared product data.
Expected output: A native block feasibility review
Review: Verify compatible field types and behavior with blank values.
Bring: Approved field mapping and unpublished theme
In this unpublished theme, build the smallest product accordion for the approved field mapping. Render product-specific content from the current product, keep shared policy links separate, and omit optional rows with no content. Use semantic details/summary where suitable, visible focus styles and existing theme typography. Preserve rich text safely through Shopify’s supported rendering. Do not duplicate the full product description, recreate review integrations, or add external scripts. Provide file changes, preview checks and rollback instructions. Do not publish.
Expected output: A small accordion implementation draft
Review: Test different products, empty fields, keyboard use and preserved content.
Bring: Actual bill, billing interval, build hours, labor value, maintenance and added tools
Calculate gross avoided subscription costs separately from first-year net benefit. Use my actual billing interval, renewal date, taxes and prepaid commitments; do not assume an immediate refund. Subtract setup labor, recurring maintenance and additional tool or hosting costs. Show every assumption, formula and a higher-effort scenario. Mark hypothetical inputs and ask for missing figures.
Expected output: A transparent cost comparison
Review: Recalculate totals and confirm cancellation and renewal terms.
Bring: Current app features used, official free-plan limits and theme capabilities
Compare these required functions with the supplied official free-plan limits and native theme capabilities. Identify what would be lost by downgrading. Recommend the simplest option that meets the requirements, including keeping the paid plan if needed. Do not assume a free plan has identical features or invent current pricing.
Expected output: A keep, downgrade or build comparison
Review: Verify current pricing and whether the feature limits fit the whole catalog.
Bring: App settings, exports, fields and documented integrations
List content and data owned by this app versus Shopify or another provider. Identify what may disappear or stop updating after removal. Propose a preservation and mapping checklist. Flag exports we do not have. Do not copy proprietary code, scrape other stores or uninstall anything.
Expected output: A preservation checklist
Review: Confirm exports can be read and all required records are represented.
Bring: Original used functions and candidate replacement
Create acceptance checks comparing the original used functions with this replacement. Include different products, blank optional fields, mobile widths, keyboard navigation and existing integrations. State the evidence needed for each check. Mark all checks not yet run; do not infer success from screenshots.
Expected output: A comparison checklist
Review: Execute the checks in preview and record actual failures and passes.
Bring: Approved change files, field definitions and preview results
Write a short maintenance handoff for this theme customization. Name changed files, data sources, owner, edit steps, known limits, checks after theme updates and rollback steps. Distinguish reverting theme files from restoring shared product data. Do not claim a theme duplicate backs up the entire store.
Expected output: A maintenance handoff
Review: Have the future owner follow the instructions in a safe preview.
Bring: Subscription terms, renewal date and non-sensitive billing records
Create a checklist to verify that this app replacement actually reduces the bill. Separate disabling a widget, uninstalling an app, canceling an externally billed subscription and any prepaid commitment. Identify which records prove the change. Do not claim cancellation or refunds have occurred.
Expected output: A billing verification checklist
Review: Check the app provider and Shopify billing records after the relevant billing cycle.
App replacements · General assistant + Shopify Flow editor
Plan one order-tagging rule
Bring: Exact rule, tag spelling, downstream consumers and plan
Draft a Flow plan for [event and condition] adding [exact tag]. Do not change fulfillment, inventory or existing tags. Identify editor fields I must verify. Include matching, nonmatching and missing-data cases. List downstream dependencies and exclude historical backfill unless separately specified. Do not claim installation or test success.
Expected output: A bounded configuration plan and cases
Review: Confirm fields in Flow and inspect final tags after other processes run.
Illustrative prompt — not execution-tested
App replacements · General assistant
Stress-test replacement economics
Bring: Actual billing interval, avoidable charges, setup estimate and maintenance range
Compare [app] with replacing only [function] over [period]. Separate cash charges and owner labor. Calculate break-even only when recurring savings are positive. Double maintenance in a second scenario. Account for retained features that prevent cancellation. Keep 30-day cycles distinct from calendar years and mark unknown costs.
Expected output: Base and higher-maintenance cost scenarios
Review: Recalculate arithmetic and verify charges really stop.
Illustrative prompt — not execution-tested
App replacements · General assistant
Investigate a disappearing order tag
Bring: Redacted event timeline, Flow result and other tag-writing systems
Review this timeline for an order tag that appears after Flow runs and later disappears. Separate observed events from hypotheses. Identify other writers to inspect and evidence needed to establish which update changed the tag. Do not recommend a blind delay, disable workflows or change live orders.
Expected output: An evidence-based investigation checklist
Review: Confirm the responsible writer before altering workflow timing.
Bring: Theme files, exact tag, approved label and placement
Inspect this unpublished Shopify theme. Propose the smallest change to show [label] for products with [exact tag] at [placement]. Reuse theme conventions, preserve existing controls and identify missing product context. Do not edit shared product tags or publish. Include tagged, untagged, long-title and phone checks plus removal instructions.
Expected output: A narrow theme-change proposal
Review: Check native settings first and verify every intended placement.
Bring: Public app docs, redacted configuration and required operations
Review replacing [app function]. Map persistent records, scheduled actions, customer controls, integrations and failure recovery to their current owners. State evidence needed before a replacement assumes each responsibility. Separate read-only helpers from full replacement. Mark unsupported migration paths unknown and do not recommend uninstalling yet.
Expected output: A responsibility and evidence table
Review: Have staff who handle exceptions verify requirements and ownership.
App replacements · General assistant + Search & Discovery
Review complementary product pairings
Bring: Product identifiers, specifications and approved compatibility facts
Using only this product table, propose up to three complementary items for each main product. Return identifiers, reason, supporting field and unresolved compatibility questions. Separate substitutes into a related-products table. Do not invent compatibility, purchase history or discounts. Mark missing evidence needs review. Do not edit the store. Start with ten products.
Expected output: A product-pairing review sheet
Review: Verify compatibility and theme support; inspect the actual storefront before removing an app.
App replacements · General assistant + Shopify Flow
Plan a low-stock threshold alert
Bring: Single-location stock rule, threshold and fixed staff recipient
Plan a Shopify Flow alert when tracked variant inventory moves from at least 5 to below 5. Verify current trigger fields in the editor; do not invent workflow JSON or enable anything. Use a fixed staff recipient. Test 6-to-4, 4-to-3, 3-to-8, 8-to-2 and already-low items. Flag missing values and multi-location limitations. Do not place orders, change inventory or contact customers.
Expected output: A bounded workflow plan and acceptance table
Review: Test in an authorized setup; verify run history and receipt, then assign ongoing ownership.
Planning · General assistant + Search & Discovery reports
Diagnose searches with no results
Bring: Redacted search queries, report dates and catalog facts
Classify these queries as possible naming mismatch, visibility issue, missing assortment, ambiguous intent or unknown. Return query, candidate product identifier, supporting evidence, next manual check and a justified fix. Separate no-results and no-clicks cases. Do not invent products, demand, synonyms or availability. Do not edit the store. Prioritize ten rows and explain why.
Expected output: An evidence-based search issue list
Review: Reproduce queries in the intended market and verify current admin controls.
Content · Sidekick product editor or general assistant
Draft a description from verified facts
Bring: Approved facts, included items, exclusions and brand voice
Draft a short product opening and features list using only the approved fact sheet. Preserve measurements and distinguish included items from accessories. Do not invent benefits, certifications, compatibility, care or shipping promises. Put missing facts in a separate review list and return a claim-to-source table. Do not publish or change other product fields.
Expected output: A draft plus separate review notes
Review: Remove working notes before saving. Verify claims and implied benefits; preserve the old description.
Bring: Source text, candidate translation, target locale and approved glossary
Compare the source and candidate translation using only the approved fact sheet and glossary. Return source phrase, translated phrase, meaning change, number or unit mismatch, glossary conflict and question for a bilingual reviewer. Separate uncertain judgments from evidenced discrepancies. Do not invent policies, certify accuracy or publish changes.
Expected output: A review queue for a fluent reviewer
Review: Check all supported and unsupported storefront surfaces; human approval remains necessary.
Development · Shopify theme editor block generation
Generate a simple theme feature block
Bring: Eligible theme, approved content, images and unpublished duplicate
Create a three-card product feature block with editable headings, body, optional image and alt text. Use theme typography and colors. Stack on narrow screens and allow long text without clipping. Hide absent images. Render optional links only with both label and URL. Use semantic headings and visible focus. No external requests, tracking, animation, product edits, cart changes or checkout changes. Explain settings and remaining tests.
Expected output: An initial theme block to inspect
Review: Check eligibility, empty fields, long text, keyboard links and multiple instances before publishing.
Bring: Untouched original, candidate, allowed fields and identifiers
Compare both CSVs read-only using a CSV parser. Report headers, row counts and matching rule. Stop on ambiguous identities; do not assume SKU is unique. List every added/deleted column or record and changed value, with identifier, field, old value, new value and whether allowed. Flag populated cells becoming blank and variant/image relationships changing. Do not fix, normalize or import. Report incomplete comparisons.
Expected output: A complete change report or an explicit comparison failure
Review: Independently check counts and identities; approve a representative small import with a recovery plan.
Bring: Audience, real content promise, cadence, incentive and configured subscription process
Draft three signup options with title up to 50 characters, content up to 100 and button up to 25. Count characters including spaces. Use only the supplied incentive and cadence. Do not invent benefits, deadlines, discounts or subscriber counts. Write a success message that matches the actual confirmation process. List settings a person must verify. Do not publish or send email.
Expected output: Three copy options and a configuration review list
Review: Independently count characters; test consent, existing customers, confirmation and unsubscribe.
Content · Shopify file editor, subject to current eligibility
Edit a background without changing the product
Bring: An authorized product photo and preservation checklist
Place the photographed product on a neutral light-gray studio surface. Preserve exact shape, proportions, color, texture, label and visible details. Use a soft contact shadow. Do not add props, accessories, packaging, text or products. Do not remove marks or wear relevant to the item sold. Keep the product fully visible with space around its outline.
Expected output: An image candidate for fidelity review
Review: Untested prompt; compare original and candidate at actual pixels and storefront size. Reject changed product details.
Planning · Shopify Inbox persona editor / Sidekick refinement
Draft a bounded Inbox agent persona
Bring: Service style and separately verified store sources and handoff settings
Use short clear sentences and a warm matter-of-fact tone. Acknowledge frustration. Explain options without purchase pressure. Distinguish documented facts from uncertainty. Do not invent benefits, dates, discounts or policy exceptions. Respect the configured handoff process. Do not claim actions completed unless they actually completed. Keep facts in approved store sources rather than this persona.
Expected output: A short persona draft for review
Review: A prompt is not a guarantee. Test policy exceptions, unsupported promises, discount requests and human handoff.
Bring: Audience definition, cutoff, subscription requirement and synthetic boundary cases
Draft a segment for [audience] with the exact cutoff [boundary]. Include only current marketing subscribers. Define behavior for customers without orders. Explain each filter and missing-value behavior. Return matching and nonmatching cases. Do not send campaigns or change consent.
Expected output: A query brief with acceptance cases
Review: Validate current syntax, actual sample membership and sending eligibility separately.
Bring: Sanitized physical-return table with row IDs, units, reasons and interval
Preserve source reasons and row IDs. Total physical returned units by SKU and reason. Keep unknown reasons separate. Propose three investigative questions with supporting rows and disconfirming evidence. Do not infer intent or calculate rates without a valid denominator.
Expected output: Reconciled totals and bounded hypotheses
Review: Recalculate totals and distinguish physical returns, reversals and refunds.
Bring: Expected behavior, sanitized trigger/branch/action evidence and missing information
Diagnose the first mismatch between expected behavior and this run. Separate facts from possible causes. Propose the smallest change and matching, nonmatching and missing-data test cases. Do not activate or rerun anything. Explain whether repeated actions could have consequences.
Expected output: A focused diagnosis and test plan
Review: Verify the resource, step data and destination result; review repeat effects before retrying.
Bring: Verified catalog fields, representative values and shopping tasks
Map each customer question to an existing catalog field, display label and test case. Flag missing facts and excessive values. Propose spelling merges without losing original values. Do not infer product facts, edit products or publish changes.
Expected output: A field map and storefront test cases
Review: Check technical distinctions, actual theme support and known-product results.
Bring: Exact component variants, quantities, choices and operational requirements
Classify this offer as fixed, multipack or mix-and-match. List component variants and quantities. Flag subscriptions and custom items. Separate presentation from inventory, checkout and fulfillment. Draft available, scarce-component and partial-return test cases. Do not modify the store.
Expected output: An offer specification and acceptance brief
Review: Verify native compatibility, component settings and actual total costs before canceling an app.
Bring: Sanitized review IDs, product, rating, date and source text
Preserve review IDs and original text. Assign zero or more themes; retain positive, negative and contradictory evidence. Count distinct reviews per theme and state the sample size. Leave unclear comments unclassified. Suggest questions, not causes or invented testimonials. Do not change ratings, verification or publication status.
Expected output: Traceable themes and investigative questions
Review: Read supporting rows, check negation and overlapping counts, and retain the original source.
Before You Switch Review Apps, Audit Your Judge.me Export
Bring: Anonymized status counts, product keys and destination fields
Given only anonymized row labels, product keys, moderation states and destination field names, reconcile counts by state. Flag unmatched products, ambiguous mappings and statuses the destination cannot represent. Preserve archived rows separately. Do not infer verification or produce an upload file.
Expected output: Reconcile the review count, moderation status and product mapping before you let a new widget display customer reviews.
Review: Before disconnecting the old app, compare the destination preview against one published, hidden, pending and archived example. Confirm that reply text stays attached to the right review and media remains available. Strip emails, IP addresses and customer details from the AI working copy. A failed status check is a reason to hold the import even if the total number of rows matches.
Group these sanitized barcode strings by exact value. Retain leading zeros. List distinct variant IDs and SKUs for each string. Separate repeated rows from codes shared by different variants. Do not generate replacement identifiers. Return only conflicts and the evidence needed to resolve each.
Expected output: Trace a duplicate scan to the physical product and variant before changing a barcode or paying for catalog cleanup.
Review: Use a test cart to scan each affected item after an authorized correction. Confirm the selected size and color, then remove the test items without completing a sale. A one-time worksheet can replace an occasional duplicate report. It cannot monitor supplier feeds. If an app also runs continuous alerts, include the cost of repeating the check before dropping it.
Bring: Fictional totals, amount paid and fulfillment state
Using these fictional order totals, paid amount, fulfillment state and requested swap, calculate the difference and identify who owes whom. List assumptions about shipping, tax and discounts. Flag missing inputs and separate payment actions from warehouse instructions. Do not edit an order, issue a refund or contact the buyer.
Expected output: Review a requested swap, check shipping and discounts, and distinguish an updated order total from money collected or refunded.
Review: A useful acceptance test includes an increase, a decrease and an even exchange. For each, verify the exact variant the warehouse sees and the payment status the agent sees. A staff checklist can cover low-volume requests. If you pay an app for buyer self-service or to synchronize warehouse changes, this manual workflow leaves those jobs with staff. At eight edits a week and five extra minutes each, that is roughly 35 hours per year before exception handling.
Shopify Color Swatches: When Can You Drop the Paid App?
Bring: Choose Sand / Small and the merchant inputs listed in the guide
Using this list of swatch placements, variant examples, staff rate and app invoice, build a keep-or-cancel worksheet. Give each placement a keyboard, sold-out and cart check. Calculate first-year savings after setup and recurring upkeep. Mark missing inputs unknown.
Expected output: Compare a native variant picker with Color Swatch King before the next renewal.
Review: Check the buying path for one representative shirt.
Audit Shopify Unit Prices Before a Multipack Promotion
Bring: $9 / 250 g and the merchant inputs listed in the guide
Using these fictional SKU, pack count, net contents, unit and price rows, normalize comparable packs to grams and calculate price per kilogram. Show each denominator. Flag missing quantities and mixed gift boxes for human review. Never infer a net weight from shipping weight.
Expected output: Catch a pack-size denominator error before it makes a six-pack look more expensive than a single bag.
Review: The denominator is the contents sold in one purchase.
Judge.me Referrals: What Does a Two-Sided Reward Cost?
Bring: $50 less $5 discount and the merchant inputs listed in the guide
Compare one-way and two-way offers using these fictional basket, variable cost, friend discount, advocate discount and redemption inputs. Show incremental contribution per additional order and the break-even order count. Flag refund lag, fixed costs and repeat orders that might have happened anyway.
Expected output: Model a friend discount, an advocate reward and the repeat purchases needed to cover the difference.
Review: Illustrative cohort. Repeat-order contribution depends on incrementality.
Why Did Pattern Swatches Disappear from Your Shopify Collection Filter?
Bring: Gingham and the merchant inputs listed in the guide
Given a filter source, two pattern entries, expected products and observations from two themes, group failures by assignment, visual data or rendering. Give the smallest next check for each. Keep missing evidence unknown and make no catalog edits.
Expected output: Trace a missing pattern image from its catalog entry to the filter drawer after a theme change.
Review: Keep one solid color as a control beside the pattern.
Replacing a Shopify Download: What Do Previous Buyers Receive?
Bring: Original link and the merchant inputs listed in the guide
Given fictional filenames, variant names and buyer entitlements, draft a release worksheet for a correction versus a paid edition. Specify expected results from old and new receipts and an internal version marker. Flag ambiguous entitlements. Do not send email or use real download URLs.
Expected output: Choose a file name and delivery plan for a correction, a new edition or a revoked download.
Review: A downloaded local copy still needs to be replaced by its owner.
Why Shopify Shows 24 Units on Hand but Only 11 Available
Bring: 24 cups counted and the merchant inputs listed in the guide
Using fictional counts for one SKU and location, reconcile on hand against available, committed and unavailable. List incoming separately. Model releasing only the inspected quantity, show before and after totals, and flag unexplained differences. Never post inventory adjustments.
Expected output: Reconcile a quality hold at one location without accidentally offering damaged or committed stock for sale.
Review: An incoming carton belongs on a separate line until received.
Save your tool, prompt, inputs, revisions and observed results. A useful prompt library gets better through documented outcomes. Copying a prompt here does not prove an app runs correctly or that an app subscription can be canceled.