API Automation
RESTful Booker API Automation
REST API Testing Framework with Playwright & TypeScript
An independent API test automation case study using Playwright and TypeScript to test the public RESTful Booker demo API: authentication, booking create, read, update and delete, negative cases and multi-step workflows, run in GitHub Actions.
01 Context
Project Context
This is an independent, public QA automation project built to show how I structure and run API tests. The target is the RESTful Booker API, a public booking system designed for API testing practice.
The project prioritizes code quality, maintainability, and clarity. It demonstrates how to design reusable API clients, organize test suites effectively, and structure meaningful test coverage rather than simply increasing test count.
02 System under test
API Testing Workflow
The test suite validates a complete API lifecycle:
This flow validates state management, data persistence, and the correctness of multi-step API interactions. Each step verifies response structure, data integrity, and error handling.
03 Approach
An API client method, from the repository
async post<T>(endpoint: string, payload: RequestPayload, token?: string): Promise<T> {
const headers = this.getHeaders(token);
const response = await this.request.post(`${this.baseURL}${endpoint}`, {
headers,
data: payload,
});
return response.json();
}
// ...
if (token) {
headers['Cookie'] = `token=${token}`;
}
createBooking() is this.post('/booking', booking). src/api/clients/api-client.ts, lines 17–24, 86–88 @ a385240 (opens in a new tab)04 Coverage
Test Coverage
47 tests organized across 9 suites covering distinct API concerns:
05 HTTP methods
HTTP Methods Coverage
The tests send GET, POST, PUT, PATCH and DELETE requests. They pin the status codes RESTful Booker actually returns, listed under Testing Strategy below. Some are not the textbook code: a PUT or DELETE on a booking that does not exist gets 405, and a failed login gets 200.
- GET
- Retrieve booking data
- POST
- Create new bookings
- PUT
- Replace booking entirely
- PATCH
- Partial booking updates
- DELETE
- Remove booking
06 Strategy
Testing Strategy
- Status Code Validation
-
22 of the 47 tests assert a status code. These are all the codes asserted, by request, with the number of tests that assert each:
GET /ping: 201 (1 test)POST /auth: 200 (3 tests)POST /booking: 200 (1 test), 500 (2 tests)GET /booking/{id}: 404 (6 tests)PUT /booking/{id}: 400 (1 test), 403 (2 tests), 405 (1 test)PATCH /booking/{id}: 403 (1 test)DELETE /booking/{id}: 403 (2 tests), 405 (2 tests)
No test asserts a 401. The status of a successful create, update, patch or delete is not asserted either, because the client helpers that return a parsed body do not check it. The 200 on
POST /authis a failed login (the body carries a reason), and the 500 is a booking missing a required field. - Response Structure
Validate JSON payload structure, data types, and required fields
- Data Integrity
Confirm sent data matches retrieved data and persists across requests
- Authentication
Test token creation, validation, and rejection of invalid credentials
- Negative Testing
Validate API behavior with invalid inputs, edge cases, and error scenarios
- Workflow Validation
Test multi-step sequences and state management across requests
07 Architecture
Framework Architecture
The layout of the repository at the pinned commit (a385240): client classes, models and test data under src/api, and one spec file per area under tests/api.
src/api/
├── clients/
│ ├── api-client.ts (base class)
│ ├── auth-client.ts
│ └── booking-client.ts
├── models/
│ └── booking.types.ts
├── test-data/
│ └── booking-data.ts
└── config.ts
tests/api/
├── auth.spec.ts
├── booking-create.spec.ts
├── booking-delete.spec.ts
├── booking-partial-update.spec.ts
├── booking-read.spec.ts
├── booking-update.spec.ts
├── health-check.spec.ts
├── negative-testing.spec.ts
└── workflow.spec.ts
playwright.config.ts
.github/workflows/playwright.yml
Key Design Principles:
-
Reusable API Clients — A base
APIClientholds the HTTP calls and headers;AuthClientandBookingClientadd one named method per endpoint. Four tests send a raw request instead: the health check, two malformed booking IDs and a PATCH without a token. -
TypeScript Interfaces —
booking.types.tsdefines interfaces for a booking and the two responses, which type the test data and the client methods. They are checked at compile time only; nothing validates a response against them at run time. -
Centralized Test Data —
src/api/test-data/booking-data.tsexports four sample bookings. The specs use three of them (validBooking,updatedBookingandbookingWithoutAdditionalNeeds);validBooking2is not used. There are no Playwright fixtures: each test creates its own bookings. -
Clean Test Organization — One spec file per operation (create, read, update, partial update, delete), plus health check, authentication, negative cases and workflows: nine files under
tests/api. - Assertions — Status codes (22 of the 47 tests), response fields and a read-back after writes. Three negative tests cannot fail and the delete test checks only that the reply is truthy (see Negative Testing below).
08 Negative tests
Negative Testing
The test suite includes 8 negative tests, covering these areas:
Two of the eight assert a rejection (an incomplete PUT returns 400, a non-numeric ID returns 404). The other six record
input this practice API accepts: zero and negative prices, an invalid date, a checkout before the check-in,
special characters in the login and a 1,000-character name. Three of those six use
expect(response.ok).toBeTruthy() without calling ok(), so the assertion cannot fail.
The article explains what I would change.
09 Skills
What This Project Demonstrates
- API Testing
- TypeScript
- Test Architecture
- Negative Testing
- CI/CD Automation
10 Evidence
Evidence & Verification
Source code, tests and the CI workflow are public on GitHub. CI run logs and report artifacts need a GitHub sign-in. Artifacts are kept for 30 days, so those from the 10 September runs have expired.
-
Source Code
The Playwright Test project: API client classes, models, test data and the 47 test definitions.
-
CI/CD Execution
GitHub Actions workflow that runs the tests on every push and pull request to main. Its run history has 3 runs: two on 10 September 2026 (the first failed, the second passed) and one on 22 October 2026 (passed; by my reading of its job log, no retry). Job logs require a GitHub sign-in.
-
Documentation
README file with project structure, test descriptions, setup instructions, and framework architecture overview.
Explore the Project
Inspect the complete source code, test suites, CI/CD configuration, and documentation on GitHub.
For the decisions and weaknesses behind these tests, read API Testing Lessons from a Playwright and TypeScript Suite. To see which API risks these tests do and do not cover, next to my other two repositories, see the API testing evidence map. If you need similar API testing, see QA Services.