QA Office

Multi-agent AI-powered QA automation platform. Upload test cases or generate them from requirements — the pipeline executes them against your app using Playwright and produces an Excel report.

Architecture

15-agent pipeline: A0 (generate) → A13 (env check) → A14 (POM build) → A10 (plan) → A12 (seed) → A1 (ingest) → A2 (intent) → A15 (validate) → A3 (locator) → A4 (execute) → A5/A6 (diagnose/heal) → A8 (report) → A10 (reflect) → Synthesis

Quick Start (Docker)

cp .env.example .env          # fill in ANTHROPIC_API_KEY, OPENAI_API_KEY
docker-compose up --build

Frontend: http://localhost:3005 — Backend: http://localhost:8005

Quick Start (Local)

Backend:

cd backend
pip install -r requirements.txt
playwright install chromium
python main.py

Frontend:

cd frontend
cp .env.example .env.local    # set BACKEND_URL=http://127.0.0.1:8005
npm install
npm run dev

Environment Variables

Variable	Required	Description
ANTHROPIC_API_KEY	Yes	Claude API key (agents A2–A15)
OPENAI_API_KEY	Yes for scriptless/scripted modes	GPT key for Playwright script generation
CORS_ORIGINS	No	Comma-separated allowed origins (default: localhost:3005)
BACKEND_URL	Frontend only	Backend URL seen by Next.js server (default: http://127.0.0.1:8005)

Execution Modes

Mode	Description	Requires
page_check	Checks page loads and content length	Nothing extra
scriptless	AI generates Playwright steps per test	OpenAI key
scripted	AI generates full Playwright script	OpenAI key