In command center, implement prompt instruction alongside url

The Churninator

The Autonomous AI Mystery Shopper for SaaS.
Find your competitor's fatal flaw before they do.

---

Churninator is an open-source AI platform that autonomously analyzes the signup and onboarding experience of any web application. It acts as a tireless "mystery shopper," navigating complex user flows to identify friction points, conversion killers, and usability issues that cause users to churn.

The output isn't just a log; it's a premium, AI-generated UX audit, complete with visual "before and after" mockups, providing product teams with the actionable intelligence they need to win their market.

✨ Key Features

• 🤖 Autonomous Agent: Powered by a fine-tuned multimodal AI that understands and interacts with modern web apps like a human.
• 📺 Live Observatory: Watch the agent's browser view and "inner monologue" in real-time as it executes its mission.
• 📄 AI-Powered Friction Reports: Receive detailed, shareable PDF reports that don't just identify problems—they propose and visualize solutions with AI-generated design mockups.
• 🔧 The Forge: A complete, open-source ML pipeline to fine-tune your own specialized agents for any task.
• ☁️ SaaS & Self-Hosted: Use our upcoming managed platform, Churninator Cloud, or self-host the entire system for maximum control.

🏗️ Architecture Overview

Churninator is a production-grade, multi-component system designed for scale and reliability.

1. Frontend (web/): A Next.js dashboard for creating and observing agent runs. It features real-time log streaming (SSE) and video (MJPEG).
2. Backend (backend/): A high-performance FastAPI application that serves the API, manages the database, and queues jobs with Redis.
3. Worker (backend/worker/): Scalable Dramatiq workers that run Playwright in a headless environment to perform the browser automation.
4. Inference Server (backend/inference/): A dedicated server that hosts the fine-tuned SmolVLM model, acting as the remote "brain" for the workers.
5. The Forge (forge/): A suite of Python scripts for our two-stage model fine-tuning process, enabling the creation of new, powerful agents.

🚀 Getting Started

You can run the entire Churninator platform locally for development and testing.

Prerequisites

• Python 3.11+
• Node.js 18+ & npm
• Postgres.app (or another local PostgreSQL server)
• A Google Gemini API Key

Local Development Setup

1. Clone the repository:

   git clone https://github.com/MR-GREEN1337/churninator.git
   cd churninator

2. Configure Environment:

   • Copy the example environment file: cp .env.example .env
   • Set up a local PostgreSQL database (see instructions here).
   • Fill in your DATABASE_URL and GOOGLE_API_KEY in the .env file.

3. Run the automated setup script: This will create a Python virtual environment, install all dependencies for the backend and frontend, and download the necessary Playwright browser.

   make -f Makefile.local setup

4. Run the application: This will start all four services (frontend, API, worker, inference) concurrently using honcho.

   make -f Makefile.local run

   You can now access the dashboard at http://localhost:3000.

☁️ Churninator Cloud

While the entire Churninator platform is open-source and free to self-host, we are building a managed Churninator Cloud platform that provides:

• One-click agent deployment without any setup.
• Team collaboration features and shared reports.
• Access to our continuously improved, state-of-the-art AI models.
• Enterprise-grade security and support.

➡️ Sign up for the private beta

---

🤝 Contributing

We are building Churninator in the open and welcome contributions of all kinds! Whether you're a developer, a designer, or a UX enthusiast, we'd love your help.

• Check out our Contribution Guide to get started.
• View our public roadmap to see what's next.
• Submit bug reports and feature requests on the Issues page.

📜 License

This project is licensed under the MIT License - see the LICENSE file for details.