AI-powered test automation platform

Vision-based testing with natural language — no code required

---

About BugBuster

BugBuster is a test automation platform powered by Vision Language Models (VLM) that sees the UI like a user and understands natural language. Write test cases in plain language, without selectors, DOM knowledge, or code.

Stack	Extensibility
VLM + Grounding + Playwright	Support for other frameworks; mobile testing roadmap

Key Features

• No coding skills required — describe test steps in natural language
• Free-form test cases — no rigid structure like Gherkin, no extra abstraction layers
• No selectors or locators — no DOM access, no maintenance when layout changes
• Stable interaction with dynamic UI elements through visual perception
• Natural language steps — use verbs like click, type, scroll, hover, wait

Who It's For

Audience	Use Case
QA specialists	Fast automation without long setup; easier test maintenance
Teams without QA Automation	Introduce automated testing without dedicated automation engineers
Non-technical users	Automate scenarios without programming knowledge

---

Quick Start

Prerequisites: Git LFS

This repository uses Git LFS for Playwright binaries and related files. Without Git LFS, these files may be missing and related functionality will not work.

1. Install Git LFS (one time per machine)

Follow the official guide for your OS: git-lfs.com

Windows (Chocolatey):

choco install git-lfs
git lfs install

If Git LFS is already installed, enable it in this repo:

git lfs install

2. Clone the repository

git clone <THIS_REPOSITORY_URL>
cd bugbuster
git lfs pull

---

Installation Guide

Step 1 — Infrastructure

Create network and start infrastructure

1. Create Docker network:

docker network create bugbuster

2. Start MinIO, PostgreSQL, Redis, RabbitMQ, ClickHouse:

docker compose -p infrastructure -f infra/docker-compose.infrastructure.yml --env-file infra/infrastructure.env.example up -d

3. Configure MinIO — open http://localhost:9001 and create buckets:

   • happypass, langfuse, run-cases, backend-files, backend-images

   

4. Create databases:

docker exec -it postgres psql -U postgres -c "CREATE DATABASE portal WITH ENCODING 'UTF8';"
docker exec -it postgres psql -U postgres -c "CREATE DATABASE langfuse WITH ENCODING 'UTF8';"

Step 2 — Langfuse

Install and configure Langfuse

1. Start Langfuse:

docker compose -p langfuse -f infra/docker-compose.langfuse.yml --env-file infra/langfuse.env.example up -d

2. Configure — open http://localhost:3300:

   • Register an account
   • Create an organization (e.g., bugbuster)
   • Create two projects: clicker and rewriter
   • Generate API keys for both projects
   • Add keys to services.env.example

   

Step 3 — Services

Build and run application services

1. Configure infra/services.env.example:

   • Add Langfuse API keys: LANGFUSE_CLICKER_PUBLIC_KEY, LANGFUSE_CLICKER_SECRET_KEY, LANGFUSE_REWRITER_PUBLIC_KEY, LANGFUSE_REWRITER_SECRET_KEY
   • Add OPENROUTER_API_KEY: INFERENCE_API_KEY, SOP_REWRITER_API_KEY

2. Build images:

docker compose -p services -f infra/docker-compose.services.yml --env-file infra/services.env.example build

3. Run migrations (one-time):

docker compose -p services -f infra/docker-compose.services.yml --env-file infra/services.env.example run --rm alembic upgrade head

4. Start services:

docker compose -p services -f infra/docker-compose.services.yml --env-file infra/services.env.example up -d

---

Main Endpoints

After startup, the platform is available at:

Service	URL
Platform (Frontend)	http://localhost:3000
Langfuse UI	http://localhost:3300
Backend API (Swagger)	http://localhost:7665/docs
MinIO Console	http://localhost:9001
Playwright Trace Viewer	http://localhost:3209

Ports are defined in infra/services.env.example and docker-compose.*.yml. Adjust addresses above if you change them.

---

Next Steps

Create your first test case

Coming soon — TBD