Shopify Sidekick for Creating Custom Apps (2026)

What Sidekick Custom App Generation Actually Is

If you have ever stared at a Shopify admin screen and thought, "I just need a tiny app that hides certain products from wholesale customers," you now have a new option that did not exist twelve months ago. Starting with the Winter 2026 edition, Shopify Sidekick can scaffold custom apps directly from a natural-language prompt — no CLI install, no ngrok tunnel, no OAuth handshake, no starter template to clone. You describe the app, Sidekick writes the code, deploys it to your store as a custom app, and hands you a working admin surface.

This is a genuine shift. For years, "custom app" on Shopify meant hiring a developer, learning Remix, or at least spinning up shopify app init on your laptop. Sidekick collapses that path into a conversation inside the admin. It is not going to replace your agency for a full B2B portal, but for the small, specific, internal tools that merchants have always wanted and never quite gotten around to building, it is close to magic — when it works.

In this guide, we will walk through exactly what Sidekick's app generator builds, which plan tiers unlock it, how the workflow actually plays out step by step, a full example of a shipping-rule app generated end to end, and the hard limits that push you back toward Shopify CLI or an agent like Claude Code. If you are weighing this against traditional development, you will also find a cost breakdown based on real build hours. For broader context on where Sidekick fits in Shopify's AI stack, see our overview of what Sidekick can actually do and our deep dive on Sidekick's limitations.

Plan-Tier Requirements: Who Can Actually Use This

Sidekick's custom-app generation is gated by plan, and the gating is stricter than most Sidekick features. Writing a weekly sales summary with Sidekick is free on every plan. Generating a custom app that runs against the Admin API is not.

Here is how availability breaks down in the Winter 2026 release:

A few things worth flagging:

• Grow is the floor. If you are on Basic and you want a custom app generated by Sidekick, you will be prompted to upgrade. The Shopify plans page reflects this in the current feature matrix.
• The usage cap is soft. You can delete or archive generated apps to free up slots. The cap exists to stop you from accidentally spinning up twenty half-working experiments.
• Embedded pages count per app, not per store. An app with two admin screens uses two of your allotted embedded pages for that app, not two of three total.
• Preview status on Grow means Shopify is still tuning the code-generation model for the lower tier. Expect occasional feature gaps — things like Flow triggers and checkout extensions are Plus-and-up for now.

If you are not sure which plan makes sense for your use case, we walk through the tradeoffs in our Shopify Plus coverage, and the broader apps and integrations category covers where custom apps fit alongside App Store apps.

What Kinds of Apps Sidekick Actually Builds Well

Sidekick is not a general-purpose code generator. It is a Shopify-specialized agent trained heavily on the Admin API, GraphQL schema, Polaris UI components, and common merchant workflows. That specialization is its superpower — and its ceiling.

Apps it builds well

These are the shapes Sidekick produces reliably on the first or second prompt:

• Admin-surface utilities. Bulk editing tools, custom product filters, internal dashboards. Anything that reads data from the Admin API and presents it in a Polaris table with filters and actions.
• Webhook-driven automations. "When an order is created with this tag, post to Slack." "When a customer hits $1,000 lifetime spend, add them to a segment." These are mostly plumbing, and Sidekick handles plumbing well.
• Metafield managers. Custom UIs for editing structured metafields on products, customers, or orders. This is a category where Sidekick frankly generates cleaner code than most junior developers, because the Polaris patterns are well-documented.
• Reporting add-ons. Pull data from the GraphQL Admin API, aggregate, display. For deeper analytics work, compare against dedicated tools — our analytics and data resources cover where custom reporting beats off-the-shelf.
• Simple discount logic. Tag-based discount gating, volume tiers, customer-segment pricing. It will not generate a Shopify Functions app for checkout customization yet, but admin-facing discount managers work.

Apps it struggles with

• Storefront extensions. Theme app extensions, checkout extensions, and anything that has to render in the customer-facing storefront requires Liquid, Hydrogen, or UI Extensions knowledge that Sidekick handles inconsistently.
• Complex state machines. Multi-step workflows with branching logic and long-running jobs tend to produce code that works on the happy path and falls apart on edge cases.
• Apps requiring third-party OAuth. If your app needs to authenticate against an external service (Stripe Connect, a 3PL API, a CRM), Sidekick will scaffold placeholder code but expects you to fill in the real credential handling.
• Performance-sensitive batch jobs. Anything that has to process 100,000+ records needs proper queue and rate-limit handling, which Sidekick does not implement by default.

The general heuristic: if a solid junior developer could build it in a week, Sidekick can probably build it in an hour. If it would take a senior developer three weeks, stick with your agency.

Example Walkthrough: Building a Wholesale Visibility App

Let us make this concrete. One of the most common internal tools merchants ask us about is a wholesale-gating app — something that hides certain products from retail customers and shows them only to tagged wholesale accounts. Here is what the Sidekick build looks like end to end.

Step 1 — Open Sidekick and describe the app.

In the admin, open the Sidekick panel and type a prompt like:

"Build me a custom app that lets me tag products as 'wholesale-only' and hides them from customers who don't have the 'wholesale' tag. Include an admin page where I can see all wholesale products and toggle the tag in bulk."

Sidekick will confirm scope, then ask clarifying questions. Typical ones: "Do you want this to hide from search and collections, or only from product pages?" and "Should wholesale pricing be handled in this app or a separate one?" Answer them.

Step 2 — Review the generated plan.

Sidekick returns an app plan before writing any code. For this example, the plan usually looks like:

1. Create a custom app with Admin API scopes: read_products, write_products, read_customers.
2. Add an embedded admin page at /app/wholesale with a filterable product table.
3. Subscribe to products/create and products/update webhooks to keep metafields in sync.
4. Add a Shopify Function (on Plus) or a Liquid theme snippet (on Grow/Advanced) to gate visibility at the storefront.

This is your checkpoint. If the plan is wrong, fix it here before any code gets generated. This is where merchants who treat Sidekick like a vending machine get burned — the plan matters.

Step 3 — Generate and deploy.

Approve the plan and Sidekick generates the app, installs it on your store, and drops you into the new admin page. The whole deployment runs in Shopify's managed environment, so there is no separate hosting bill.

Step 4 — Test against real data.

This is the step everyone skips. Create a test wholesale customer, tag them, log in as that customer in an incognito window, and confirm the gating works. Also test the inverse: log in as a retail customer and confirm you cannot see the products via direct URL, collection, or search.

Step 5 — Iterate with Sidekick.

"Add a bulk tagging action." "Add a CSV export." "Log every toggle to a metafield audit trail." Each request is a follow-up prompt, and Sidekick modifies the existing app rather than generating a new one.

For a broader view of how AI agents are reshaping store operations like this, see our coverage of agentic commerce and automating a Shopify store with AI.

Limitations vs Shopify CLI and Claude Code

Sidekick's app generator is not trying to replace Shopify CLI or a general-purpose coding agent like Claude Code. It is solving a different problem: low-friction custom apps for merchants who do not have a developer. Here is how the three stack up when you are actually building something non-trivial.

Three practical implications:

• If you are a solo merchant with no dev team, Sidekick wins by a mile. You do not need git, you do not need npm, and the app lives where you work every day. See our guide to AI tools for solo Shopify store owners for the broader toolkit.
• If you have a developer on retainer, Shopify CLI with Shopify's official app development docs is still the right path. You keep code ownership, you can test locally, and you can deploy to infrastructure you control. The Remix framework that Shopify's scaffolded app template uses is stable and well-documented.
• If you are building something novel, Claude Code (or similar agent tools) paired with Shopify CLI is the most flexible combination. The agent writes the code, the CLI deploys it, and you keep full control. We cover how to think about this in our comparison of ChatGPT vs Claude for Shopify merchants.

Cost Comparison: Sidekick vs Agency vs DIY Development

Let us put real numbers on this. The app from the walkthrough above — a wholesale visibility tool with an admin page, metafield sync, and storefront gating — is a pretty typical custom-app request. Here is what building it costs across four paths, based on US market rates in Q1 2026 as reflected in BigCommerce's ecommerce development cost guide and Hostinger's Shopify app development rate benchmarks:

The Sidekick column looks too good to be true, and in some ways it is. You are trading four things for that price: full code ownership, portability to other stores, unlimited customization depth, and the ability to bring the app to the public App Store. If those tradeoffs do not matter for an internal tool, Sidekick is an obvious win.

A few caveats on the numbers:

• Agency pricing assumes scoping, design review, QA, and a written spec. This is why the hours go up — most of the time is not coding.
• Freelance pricing assumes you bring the spec and do your own QA. Cheaper and faster, but you are the project manager.
• DIY pricing assumes you know Remix, React, and the Admin API. If you do not, add 40-80 hours of learning curve before you write a line of production code.
• Sidekick pricing assumes the app fits inside what Sidekick can generate. When it does not, you end up paying for one of the other three paths anyway after a week of frustration.

For merchants thinking about the broader trend, Shopify's Editions hub frames Sidekick app generation as one of the flagship features of the Winter 2026 release, alongside Sidekick Pulse and the expanded agentic commerce stack.

When to Use Sidekick vs When to Skip It

After building two dozen apps with Sidekick in the past quarter, here is the heuristic we actually use in practice.

Use Sidekick when:

• The app is internal-only. If only your team will ever see it, managed-by-Shopify is a feature, not a bug.
• The data flow is read-heavy. Dashboards, reports, and filtered views are Sidekick's sweet spot.
• The logic is tag- or metafield-based. Anything that boils down to "look at this field, do this thing" generates cleanly.
• You want iteration, not perfection. Sidekick is great at v1; you can refine with follow-up prompts.
• You need it next Tuesday. Speed-to-first-version is the killer feature.

Skip Sidekick when:

• You need custom checkout. Checkout UI Extensions and Shopify Functions for checkout are better handled by a developer following Pagefly's deep dive on checkout extensibility and the underlying extension SDK directly.
• The app will be sold on the App Store. Generated apps cannot be submitted to the public App Store as of the Winter 2026 release.
• You have strict compliance requirements. HIPAA, SOC 2, or PCI workflows need code paths you can audit in detail, not managed code.
• The app is mission-critical and high-volume. If downtime costs you $10K/hour, you want deployment control.
• It is a headless or Hydrogen build. Sidekick does not generate Hydrogen code yet. See our headless and Hydrogen resources for that path.

A good smell test: if you can explain the app in two sentences and a wireframe, Sidekick will probably nail it. If explaining it takes a half-hour meeting and three follow-up emails, you want a developer.

Common Mistakes Merchants Make With Sidekick Apps

Every new tool generates a predictable set of foot-guns. Here are the ones we see weekly in merchant communities and the broader Talk Shop community.

1. Skipping the plan review. Sidekick shows you a plan before it writes code. Most merchants skim it and approve. Then they wonder why the app does not do what they wanted. Read the plan. Challenge the scopes. Ask "what happens when X is missing?" before code generation.

2. Using Sidekick for storefront logic. The generator is tuned for admin surfaces. Storefront gating, custom product pages, and theme modifications are still best done in Liquid or a theme app extension. We cover this in our theme and design guides.

3. Over-scoping in one prompt. "Build me a full inventory management system with forecasting, reorder points, supplier integrations, and an analytics dashboard." This prompt produces a mess. Break it into five apps or five iterations of one app.

4. Ignoring webhook cleanup. Generated apps subscribe to webhooks. If you delete the app later without uninstalling cleanly, you can leave orphaned subscriptions that quietly fail. Always uninstall via the admin, not by deleting the app record.

5. Treating Sidekick output as finished code. Even when it works, review the code. Check the GraphQL queries for N+1 patterns, check the error handling, check the rate-limit behavior. Sidekick is a junior developer with good taste — it still needs code review.

6. Assuming it will handle Flow integration. Shopify Flow triggers and actions from custom apps require a separate app extension declaration that Sidekick does not always generate. If you want your app to appear as a Flow trigger, say so explicitly in the prompt.

7. Forgetting about the usage cap. On Grow, you get three generated apps. Delete the experiments. Archive the ones you are not using. Otherwise you hit the cap at the worst possible moment.

8. Not documenting the prompt. When the app breaks in six months and you need to regenerate or debug, the original prompt is your spec. Save it. We see merchants lose the prompt and end up reverse-engineering their own app.

Best Practices vs Common Mistakes

The Verdict: A Genuine New Tier of Merchant Tooling

Sidekick's custom-app generator is not going to replace your development partner, and it is not supposed to. What it does do is collapse the distance between "I wish I had a little tool that did X" and "I have a little tool that does X" from weeks to hours, for a specific and useful subset of apps.

If you are a merchant on Grow, Advanced, Plus, or Enterprise and you have never commissioned a custom app because the friction was too high, this is the feature to try first. Pick an internal tool that has been on your wish list. Write a two-sentence description. Open Sidekick. See what happens. The worst outcome is you spend thirty minutes and learn something; the best outcome is you ship the tool you have wanted for a year. For a broader look at how AI is reshaping the merchant toolkit, explore our AI and emerging tech category or stop by the Talk Shop community to compare notes with other merchants who are building with Sidekick.

What internal tool have you always wanted but never commissioned? Try describing it to Sidekick this week and tell us what happened.

About Talk Shop

The Talk Shop team — insights from our community of Shopify developers, merchants, and experts.