Взаимодействие с API

Real apps rarely live alone. A weather app asks a weather service for data. A crypto tracker asks a price service. The way apps ask each other for data is called an API — an Application Programming Interface.

Think of an API like a waiter: you don't go into the kitchen, you just ask for what you want and the waiter brings it back. As a vibe coder, you mostly describe what data you need, and AI writes the code that talks to the API for you.

When your app gets data from the internet, we say it fetches it. The app sends a request to a URL, and the service sends back data — usually as JSON, a simple text format that looks like labeled boxes of values.

You don't need to memorize how this works. You just need to know the words so you can tell AI: 'fetch the data and show it on the page.'

Let's ask AI to fetch real data. The trick is to name the data you want and where to show it. Here is a prompt you could paste into Cursor, Claude Code, Bolt or v0.

Notice how the prompt names a specific free API and exactly what to display. Specific prompts get working code on the first try.

When AI writes the code, you'll often see a fetch() call. You don't have to write it by hand, but recognizing it helps you read what AI gives you.

This tiny example fetches a fake user and logs their name. Run it to see data come back from the internet.

API data comes back as JSON. To use a value, you reach into it by name with a dot, like user.name. Nested data uses more dots: user.address.city.

When you're unsure what's inside a response, ask AI: 'log the full response so I can see its shape, then show me how to read the fields I need.'

The internet isn't always reliable. A request can be slow, or the service can be down. Good apps show a loading message while waiting and a friendly error message if something breaks.

Always remind AI to handle these. A prompt like the one below produces a much more robust app than just 'fetch the data.'

You can build a lot with free, no-signup APIs. Great ones for learning:

• JSONPlaceholder — fake users, posts and todos
• Open-Meteo — real weather, no key needed
• REST Countries — country info and flags
• The Cat / Dog API — random pet pictures

Tell AI which one to use and what to display, and you have a working app in minutes.

Some services need to know who is asking, so they give you an API key — a secret password for your app. You include it with each request.

Keys keep services from being abused, but they must be kept secret. For now, just know that 'this API needs a key' means an extra setup step. We cover keeping keys safe in a later lesson.

When you have an API's documentation, you can hand it to AI directly. Paste the example URL and a sample response, and let AI map it to your screen.

This is one of the biggest vibe-coding superpowers: you don't read the docs deeply, you let AI translate them into working code.

Powerful apps mix data from several sources. A travel app might combine a country API for facts, a weather API for the forecast, and an image API for photos.

You can ask AI to combine them step by step. Build one API first, confirm it works, then say 'now also add the weather for that country.' Small steps keep the AI accurate.

There are thousands of APIs, and you'll never memorize them. The vibe-coding skill is knowing the vocabulary — fetch, request, response, JSON, key — so you can describe what you want clearly.

Whenever you hit a new API, paste its docs to AI and say what data you need and where to show it. That single habit unlocks almost any data source on the web.

You learned that apps fetch data from other services through APIs, usually getting back JSON. You saw fetch() in action, how to read response fields with dots, and why apps need loading and error states.

Most importantly: you don't memorize APIs — you describe the data you want and let AI wire it up, pasting docs and example responses when needed. Next, you'll have AI read and save files for your apps.