You're building the frontend for a store. The design is ready, but the backend isn't, and you're tired of hardcoding const products = [...] in a file.
In this tutorial we'll build a small React product page against a free mock REST API, with no backend, no API key and no signup. By the end you'll have:
- a product list with search and pagination
- each product's category and reviews loaded in the same request
- a "create product" form that shows real validation errors
- a way to test your loading and error states on demand
Disclosure: I built DevSeed API, the mock API used here. It's free, and I'd love feedback on it.
1. Look at the data
Every endpoint is a plain GET you can open in the browser:
https://www.devseedapi.com/api/v1/ecommerce/products?limit=2 { "success": true, "data": [ { "id": "p1", "name": "Wireless Mouse", "price": 29.99, "categoryId": "c1", "createdAt": "2026-03-02T10:00:00Z" }, { "id": "p2", "name": "Mechanical Keyboard", "price": 79.99, "categoryId": "c1", "createdAt": "2026-03-02T10:10:00Z" } ], "meta": { "total": 120, "page": 1, "limit": 2, "totalPages": 60 } } The e-commerce set also has categories, reviews, orders, cart and coupons, and there are user and SaaS datasets too. Every list endpoint supports the same query params:
| What | Example |
|---|---|
| Pagination | ?page=2&limit=10 |
| Search | ?search=mouse |
| Exact filter | ?categoryId=c1 |
| Range filter | ?minPrice=20&maxPrice=50 |
| Sort | ?sortBy=price&order=asc |
| Related data | ?include=category,reviews |
They combine, so ?search=mouse&maxPrice=50&sortBy=price&order=asc returns the mouse-related products under $50, cheapest first.
2. Load related data in one request
This is the part most mock APIs skip. On a real product page you need the product's category and its reviews, and you don't want to fire three requests and join them yourself.
GET https://www.devseedapi.com/api/v1/ecommerce/products/p1?include=category,reviews { "success": true, "data": { "id": "p1", "name": "Wireless Mouse", "price": 29.99, "categoryId": "c1", "category": { "id": "c1", "name": "Computer Accessories" }, "reviews": [ { "id": 101, "productId": "p1", "userId": "u1", "rating": 5, "comment": "Excellent mouse!" }, { "id": 163, "productId": "p1", "userId": "u70", "rating": 5, "comment": "Exactly as described." } ] } } (Trimmed a little for space.) include works on list endpoints too, which is what we'll use next.
3. A product list with search and pagination
Set up a React app (npm create vite@latest shop -- --template react) and add a small hook:
// useProducts.js import { useEffect, useState } from "react" const API = "https://www.devseedapi.com/api/v1/ecommerce" const LOADING = { status: "loading", products: [], meta: null, error: null } export function useProducts({ search, page, simulate = {} }) { // One string describing the request; a new value means a new fetch. const key = JSON.stringify({ search, page, ...simulate }) const [result, setResult] = useState({ key: null }) useEffect(() => { const controller = new AbortController() const params = new URLSearchParams({ limit: 8, include: "category,reviews", ...JSON.parse(key) }) fetch(`${API}/products?${params}`, { signal: controller.signal }) .then(async res => { const body = await res.json() if (!res.ok) throw new Error(body.message ?? `HTTP ${res.status}`) setResult({ key, status: "success", products: body.data, meta: body.meta, error: null }) }) .catch(err => { if (err.name !== "AbortError") { setResult({ key, status: "error", products: [], meta: null, error: err.message }) } }) return () => controller.abort() }, [key]) // Until the response for the current request arrives, we're loading. return result.key === key ? result : LOADING } And the page:
// ProductList.jsx import { useState } from "react" import { useProducts } from "./useProducts" const average = reviews => reviews.length ? (reviews.reduce((sum, r) => sum + r.rating, 0) / reviews.length).toFixed(1) : "–" export default function ProductList() { const [search, setSearch] = useState("") const [page, setPage] = useState(1) const { status, products, meta, error } = useProducts({ search, page }) return ( <div> <input placeholder="Search products" value={search} onChange={e => { setSearch(e.target.value); setPage(1) }} /> {status === "loading" && <p>Loading…</p>} {status === "error" && <p role="alert">Couldn't load products: {error}</p>} <ul> {products.map(p => ( <li key={p.id}> <strong>{p.name}</strong> · ${p.price} · {p.category?.name} · ★ {average(p.reviews)} ({p.reviews.length}) </li> ))} </ul> {meta && ( <nav> <button disabled={page <= 1} onClick={() => setPage(page - 1)}>Previous</button> <span> Page {meta.page} of {meta.totalPages} </span> <button disabled={page >= meta.totalPages} onClick={() => setPage(page + 1)}>Next</button> </nav> )} </div> ) } Each product arrives with its category and reviews already attached, so there's no second round of requests. meta.totalPages drives the pagination buttons.
4. A create form with real validation errors
Mock APIs usually accept anything you POST, so your error handling never gets exercised. Here, writes are validated like a real API but never stored: the response shows the record that would have been created.
Send a bad body:
curl -X POST https://www.devseedapi.com/api/v1/ecommerce/products \ -H "Content-Type: application/json" \ -d '{"name":"Desk Lamp","price":"cheap"}' { "success": false, "message": "Validation failed", "errors": [ { "field": "price", "message": "must be a number" }, { "field": "categoryId", "message": "is required" } ] } That's a 422. It even checks references: "categoryId": "c999" fails with no categories record with id "c999". A valid body returns 201 with a new id and a Location header.
A form that maps those errors onto fields:
// NewProductForm.jsx import { useState } from "react" const API = "https://www.devseedapi.com/api/v1/ecommerce" export default function NewProductForm() { const [form, setForm] = useState({ name: "", price: "", categoryId: "c1" }) const [fieldErrors, setFieldErrors] = useState({}) const [created, setCreated] = useState(null) async function handleSubmit(e) { e.preventDefault() setFieldErrors({}) setCreated(null) const res = await fetch(`${API}/products`, { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ ...form, price: form.price === "" ? undefined : Number(form.price) }), }) const body = await res.json() if (res.status === 422) { setFieldErrors(Object.fromEntries(body.errors.map(err => [err.field, err.message]))) } else if (res.ok) { setCreated(body.data) } } const field = name => ( <label> {name} <input value={form[name]} onChange={e => setForm({ ...form, [name]: e.target.value })} /> {fieldErrors[name] && <small style={{ color: "crimson" }}>{fieldErrors[name]}</small>} </label> ) return ( <form onSubmit={handleSubmit}> {field("name")} {field("price")} {field("categoryId")} <button>Create</button> {created && <p>Created {created.name} with id {created.id}</p>} </form> ) } Leave the price empty or type a category that doesn't exist, and you'll see the server's message under the right field. PUT, PATCH and DELETE on /products/:id work the same way.
5. Test your loading and error states
The happy path is easy. The states that break in production are the slow request and the failed one, and they're hard to trigger against a real backend. Three query params do it:
| Param | Effect |
|---|---|
delay=2000 | waits 2 seconds before responding |
error=503 | always responds with that status |
errorRate=30 | fails 30% of requests with a 500 |
Our hook already passes extra params through, so:
// See the spinner for 2 seconds useProducts({ search, page, simulate: { delay: 2000 } }) // See the error state useProducts({ search, page, simulate: { error: 503 } }) // Flaky network: does a retry button or a stale list still look right? useProducts({ search, page, simulate: { errorRate: 30 } }) The error=503 case returns { "success": false, "message": "Simulated API Error" } with a 503 status, so the component shows "Couldn't load products: Simulated API Error". This is also handy for Storybook stories and Playwright/Cypress tests of your error UI.
Swapping in the real backend later
When your backend is ready, change API to point at it. If your real API uses the same ideas (page/limit, a meta object, field-level 422 errors), the components barely change. If it doesn't, the mock gives you a concrete contract to discuss with your backend team.
A few honest limits: writes are never saved, so a POST won't appear in the next GET. The data is the same for everyone, and there's a rate limit of 200 requests per 15 minutes per client.
Wrap-up
We built a product list with search, pagination and related data in a single request, a form that handles real validation errors, and a way to test loading and failure states, all against a free API with no setup.
- Docs and a live console for every endpoint: https://www.devseedapi.com/docs/ecommerce/products
- Source: https://github.com/gayashanacd/dev-seed-api
If there's a dataset you'd want next (social media, fintech, edtech…), or something that annoyed you, tell me here. It takes under a minute, and it decides what gets built next.

