BjarniPeturFridjonsson/EndToEndTestinWithAI
E2E test infrastructure for multi-role web applications using Playwright, MCP, NodeJs, C# and Claude AI. Covers AI-assisted test authoring, automatic page coverage reporting, accessibility auditing, and role-based authentication setup. A real-world pattern, not a runnable demo. More info in the README.md
Platform-specific configuration:
{
"mcpServers": {
"EndToEndTestinWithAI": {
"command": "npx",
"args": [
"-y",
"EndToEndTestinWithAI"
]
}
}
}Add the config above to .claude/settings.json under the mcpServers key.
This repository shows how we built and maintain a practical end-to-end test suite for a real production web application — a multi-role EASA compliant flight club management system with around 110 pages across four user roles. It is not a framework you can drop into your project unchanged. It is a set of decisions and patterns you can read, and adapt.
The companion tool — TestBuilderTool/ — is a local web UI that combines Playwright browser automation with the Claude AI API to help write tests through conversation. That is documented separately in TestBuilderTool/DESIGN.md.
[](https://youtu.be/_PFQgEdFgCw)
---
AI-assisted development has fundamentally changed the speed at which code gets written. Features that used to take days take hours. That acceleration is real and valuable — but it compresses the feedback loop in ways that can bite you. When a developer and an AI pair write code quickly across many parts of a system, the risk is not that any single piece is wrong. The risk is that pieces that worked yesterday break silently when something else changes today.
End-to-end tests are the safety net for that kind of development. They do not test units in isolation — they test that the whole system, from the browser to the database, does what it is supposed to do for a real user. That is exactly the kind of regression that AI-accelerated development is most likely to introduce and least likely to catch on its own.
The problem has always been that E2E tests are expensive to write and most teams skip it or let the test suite rot. That calculus has changed. With AI assistance, the mechanical cost of writing a test drops dramatically. What used to take an hour takes minutes. That makes a comprehensive E2E suite achievable for teams that could never have justified it before.
The cr
Loading reviews...