Neutrino Test is an AI-powered QA testing platform that automates end-to-end testing using simple, natural language commands.
Make sure you have the following installed:
Navigate to the /apps/webapp directory and create an environment file (e.g., .env) with the following variables:
# Env variables for the webapp (webapp currently uses the Claude API driver) CLAUDE_API_KEY=your_claude_api_key_here SUPABASE_URL=your_supabase_url_here SUPABASE_KEY=your_supabase_key_here
Start your local Kubernetes cluster with Minikube:
Then, configure your Docker client to use the Minikube daemon:
eval $(minikube docker-env)
To verify that Docker is now using Minikube, run:
docker info | grep minikube
Navigate to the /apps/neutrino directory and build the Docker images using Docker Compose:
From the root of the repository, run the following command to start the webapp:
yarn workspace webapp dev
This command will launch the development server for the webapp
1 Comment
blckprl367
Yes, you heard that right — I open-sourced a YC company.
It was mostly out of sheer desperation. Landing a job as a recent grad these days feels harder than escaping Alcatraz.
I came across a YC company called SpurTest on the Lightcone podcast. They were working on AI-driven QA testing using natural language, and I found it fascinating.
So, I decided to build an open-source version — Neutrino.
I created an AI web agent from scratch using step prompting in a Playwright browser.
It simulates how a human would interact with a website to generate end-to-end tests, all driven by natural language.
It’s been a super fun project so far. This is just v1 — still rough in some areas, but fully working.
Right now, I'm experimenting with efficient step-mark prompting technique to make it smarter and faster.
I'm also considering fine-tuning Qwen 2.5 VL 7B for reduced cost and better performance down the line.
Come on critize, roast or even suggest