TestPilot AI 🚀

An AI-Powered REST API Test Case Generator

TestPilot AI is a full-stack web application designed to automate the creation of API test cases. By simply providing an HTTP method, endpoint, and a brief description, the application uses an LLM Service (Groq API — openai/gpt-oss-20b) to instantly generate structured positive, negative, and validation test cases.

This project was built to demonstrate backend API integration, prompt engineering, and clean, layered architecture in Java & Spring Boot.

---

🎯 Project Highlights

1. AI Integration via RESTful APIs: Instead of using bulky SDKs, the backend communicates directly with Groq's high-speed OpenAI-compatible REST API using Spring's RestTemplate.
2. Robust JSON Parsing Strategy: Large Language Models (LLMs) can sometimes return unpredictable formats (like markdown code blocks). The backend uses a custom, multi-step parsing strategy (Regex + Jackson ObjectMapper) to guarantee the frontend never crashes due to bad AI output.
3. Prompt Engineering: The AI is strictly instructed via a carefully designed system prompt to generate specific edge-cases (like 1-character boundaries, nulls, and empty fields) and output them in a strict JSON schema.
4. Clean Architecture: The Java backend strictly follows a layered architecture (Controller → Service → LLM Service layer) using Data Transfer Objects (DTOs) to decouple the AI response from the client payload.

---

⚙️ How it Works (The Architecture)

System Architecture Diagram

graph TD
    Client[React Frontend] -->|JSON Request| Controller[TestCaseController]
    Controller -->|DTO| Service[TestCaseService]
    Service -->|Prompt Builder| LLMService[LLMService]
    LLMService -->|REST API Call| API[Groq API — openai/gpt-oss-20b]
    API -->|Raw JSON Text| LLMService
    LLMService -->|Regex + Jackson Parse| Service
    Service -->|Structured DTO| Controller
    Controller -->|JSON Response| Client

Loading

Project Structure

SmartAPITester/
├── backend/                  # Spring Boot Java Application
│   ├── src/main/java/.../testpilot/
│   │   ├── controller/       # Exposes REST endpoints to React
│   │   ├── dto/              # Data Transfer Objects (Request/Response)
│   │   ├── exception/        # Global Error Handling (LLMException)
│   │   └── service/          # Business logic & AI Prompt Engineering (LLMService)
│   └── src/main/resources/   # App Config & Environment Variables
│
└── frontend/                 # React UI Application
    ├── src/
    │   ├── components/       # InputForm & ResultSection (Postman UI)
    │   ├── services/         # Axios API Client
    │   ├── App.jsx           # Main State Manager
    │   └── index.css         # Global Theme Variables
    └── package.json          # Node Dependencies

Data Flow

1. The Request: The user enters API details in the React frontend (Postman-inspired UI).
2. The Controller: The React app sends a JSON payload to the Spring Boot TestCaseController.
3. The Prompt: The TestCaseService takes the user data and injects it into a strict instruction prompt engineered for a Senior QA Engineer persona.
4. The AI Call: The LLMService sends the prompt to the Groq API (openai/gpt-oss-20b).
5. The Parsing: The backend receives a raw text response, strips away any markdown artifacts, and safely deserializes the string into Java objects (TestCaseResponse).
6. The Result: The frontend receives the clean JSON and displays it in a 3-column grid, allowing the user to download the final test suite.

---

🛠️ Tech Stack

Backend (The Core)

• Java 21
• Spring Boot (Web starter)
• Maven (Dependency management)
• Groq API (openai/gpt-oss-20b via OpenAI-compatible endpoint)
• Jackson (JSON serialization/deserialization)

Frontend (The UI)

• React 18 (Built with Vite)
• Axios (HTTP client for connecting to the backend)
• Vanilla CSS (Custom Postman-inspired dark theme)

(Note: To keep the project lightweight and focused purely on AI integration, there is no database, no authentication, and no heavy UI frameworks like Tailwind or Material UI).

---

🚀 Local Setup Instructions

Want to run this on your own machine? Follow these steps:

1. Prerequisites

• Java 21+ installed
• Node.js & npm installed
• A free Groq API Key

2. Start the Backend (Spring Boot)

Open a terminal, navigate to the backend folder, and set your API key as an environment variable:

Windows (PowerShell):

$env:GROQ_API_KEY="gsk_your_api_key_here"
mvn spring-boot:run

Mac/Linux:

export GROQ_API_KEY="gsk_your_api_key_here"
mvn spring-boot:run

The backend will start on http://localhost:8080

3. Start the Frontend (React)

Open a second terminal, navigate to the frontend folder, and run:

npm install
npm run dev

The frontend will open at http://localhost:5173

---

🔮 Live demo link:

https://test-pilot-ai-beryl.vercel.app/