SPARTAficial
Simply, Powerful, Automated, Regression, Test, Architecture.
// What Is SPARTA?
SPARTA is the test automation framework I designed at Cerios. It exists because of a pattern I kept seeing: a team builds a quick proof-of-concept, it works, stakeholders are happy, and before anyone notices the PoC becomes the production framework. Six months later it's a tangled mess that nobody fully understands, every change risks breaking something, and the team has lost confidence in their own tests.
// The Problem
PoC-to-Production
Frameworks that grew organically from a proof-of-concept. No architecture, no standards, just layers of quick fixes on top of each other.
Blind Spots
One person built the framework. That person left. Now nobody truly understands why things work, or why they suddenly don't.
Fragile Tests
A small change in the application breaks half the test suite. Not because the tests were wrong, but because the framework had no isolation or proper abstractions.
// The Solution: Layered Architecture
SPARTA solves this with a clear, layered architecture. Each layer has a specific purpose, and the boundaries between them are strict.
Building Blocks
The foundation. Page objects, API clients, component abstractions, data builders. All standardized, all following the same patterns. A new team member recognizes the structure immediately because it's consistent.
Accelerators
The open-source npm packages that speed up common tasks: test data builders, custom matchers, table assertions, step decorators. Reusable across projects, not tied to any specific client.
Guardrails
Linting rules, naming conventions, and quality gates that prevent the framework from degrading over time. The guardrails make it hard to do the wrong thing and easy to do the right thing.
// Why It Works
- Standardized: Every SPARTA project looks the same. Teams can move between projects without relearning the framework.
- AI-friendly: Clear structure and guardrails mean AI can generate tests that actually fit the architecture. No spaghetti.
- Open foundation: Built on Playwright (open source) and extended with Cerios packages (MIT license). No vendor lock-in on the core.
- Proven: Deployed at AZL, TenneT, Robeco, It&Care and De Alliantie myself. Other colleagues have successfully implemented it in their projects as well.
// SPARTAFICIAL
SPARTAFICIAL is the AI layer on top of SPARTA. It's not a separate product. It's what happens when you combine a well-structured framework with AI that actually understands that structure.
1. Introspection
The SPARTAFICIAL introspector reads your SPARTA building blocks via an API. It understands your pages, your API clients, your data builders, and your entire test vocabulary.
2. Your LLM, Your Data
Connect your own LLM API key. Your code and test logic never leave your infrastructure. SPARTAFICIAL sends the building block context to your AI provider.
3. Chatbot-Driven Generation
Describe what you want to test in natural language. SPARTAFICIAL generates tests that use your existing building blocks, not generic code that ignores your framework.
The key insight: the building blocks are made by test automators who understand code. AI uses those blocks to compose tests. This means the generated tests follow your standards, use your page objects, call your API clients, and build data through your builders, from day one.
Without structure, AI generates compilable spaghetti. With SPARTA's guardrails and building blocks, AI generates tests that belong in your codebase.