---
title: API scenarios
description: Test REST and HTTP services with API steps inside your web and mobile tests.
sidebarTitle: API
lastUpdated: "2026-09-02"
---

> **For AI agents:** the complete documentation index is at [llms.txt](/llms.txt). Append `.md` to any page URL for its markdown version.

API testing in Kapptivate covers REST and HTTP services: methods, headers, JSON bodies, response assertions, variable extraction, and detailed per-request timing. API actions live as steps inside web or mobile tests, added from the [test builder](/tests/builder) like any other step; the action and metrics are identical in both, only the surrounding test type differs.

## API actions

HTTP API actions are available inside existing tests:

- **Inside a web test**: add HTTP requests alongside browser steps. Action identifier: `ETH_HTTP_API`.
- **Inside a mobile test**: add HTTP requests alongside smartphone steps. Action identifier: `DATA_API_HTTP`.

Both share the same request configuration and response assertion model, with a few channel-specific differences in routing.

In a **Web (New experience)** test, this is the [API Call step](/tests/actions/web/apicall): same request, plus response conditions written in the step's Checks tab and a response viewer in the builder and in the results. `ETH_HTTP_API` is what you still address in classic tests and through the [CLI](/cli/guides/create-tests) and MCP.

## Request configuration

- **HTTP methods**: `GET`, `POST`, `PUT`, `PATCH`, `DELETE`.
- **URL** with templated variables.
- **Custom headers** with autocompletion for standard ones.
- **Body**: JSON or raw payload.
- **Routing**: DNS override, proxy, SSL bypass, cookie preservation, redirect handling.

## Response validation

- **Status code** checks.
- **Body** and **header** assertions.
- **Variable extraction** from the response for use in later steps.

## Per-request metrics

Each HTTP request emits a detailed timing breakdown: DNS lookup, TCP connection, TLS handshake, content generation, transfer, and total. This lets you spot performance regressions, not just functional failures.