🚀 MCP RAG Agent – AI-Powered API Testing Framework

📌 Overview

The MCP RAG Agent is an AI-driven modular testing framework that combines:

• 🔎 RAG (Retrieval Augmented Generation) – Knowledge-based context retrieval

• ⚙️ MCP Layer (Tool Execution Engine) – Executes tools dynamically

• 🧪 API Testing Agent – Automates API validation like Postman

It enables natural language → API execution → validation → intelligent response generation.

🧠 System Architecture

graph TD
A[User Query] --> B[API Agent - NLP Parser]
B --> C[MCP Server - Tool Router]
C --> D[RAG Engine - Knowledge Retrieval]
C --> E[API Execution Tool]
D --> C
E --> F[External API / System]
F --> G[Response Validation Layer]
G --> H[Final AI Response]

🧩 Architecture Explanation

1️⃣ API Agent Layer

• Accepts natural language input

• Converts request into structured API test case

2️⃣ MCP Server Layer

• Central orchestration layer

• Routes requests to appropriate tools

3️⃣ RAG Layer

• Fetches contextual knowledge from documents

• Enhances API validation logic

4️⃣ Execution Layer

• Executes API calls (GET/POST/PUT/DELETE)

• Captures response payloads

5️⃣ Validation Layer

• Compares expected vs actual response

• Returns structured test result

🔁 End-to-End Flow

User Input
   ↓
API Agent (Intent Detection)
   ↓
MCP Server (Tool Selection)
   ↓
RAG (Context Injection)
   ↓
API Execution Engine
   ↓
Response Validation
   ↓
Final Result Output

⚙️ Installation Guide

1️⃣ Clone Repository

git clone https://github.com/karthikeyanramu/MCP_RAG_AGENT.git
cd MCP_RAG_AGENT

2️⃣ Create Virtual Environment

python -m venv venv

Activate:

# Windows
venv\Scripts\activate

# Mac/Linux
source venv/bin/activate

3️⃣ Install Dependencies

pip install -r requirements.txt

4️⃣ Start MCP Server

python server/mcp_server.py

Expected:

MCP Server running on http://localhost:5000

5️⃣ Run API Agent

python -m qa_agent.api_agent_runner

🧪 Postman Integration (Manual Testing Support)

Even though this system is AI-driven, it supports Postman-style API testing.

📌 Example Request

🔹 Endpoint

POST http://localhost:5000/execute

🔹 Headers

{
  "Content-Type": "application/json",
  "Authorization": "Bearer <token-if-needed>"
}

🔹 Sample Payload

{
  "tool": "api_executor",
  "method": "POST",
  "url": "https://api.example.com/login",
  "headers": {
    "Content-Type": "application/json"
  },
  "body": {
    "username": "test_user",
    "password": "Test@123"
  }
}

📌 Sample Response

{
  "status": 200,
  "message": "Login Successful",
  "token": "eyJhbGciOiJIUzI1NiIs...",
  "validation": "PASSED"
}

🔄 CI/CD Pipeline (QA Maturity Model)

This system can be integrated into CI/CD pipelines for automated API validation.

🚀 Pipeline Flow

graph LR
A[Code Push] --> B[CI Trigger - GitHub Actions]
B --> C[Install Dependencies]
C --> D[Run API Tests via MCP Agent]
D --> E[RAG Validation Layer]
E --> F[Test Report Generation]
F --> G[Deploy / Fail Pipeline]

🧪 CI/CD Benefits

✔ Automated API regression testing ✔ AI-driven validation (reduces manual QA effort) ✔ Early defect detection ✔ Domain knowledge injection via RAG ✔ Scalable test execution

📌 Sample GitHub Actions Workflow

name: MCP API Tests

on: [push]

jobs:
  test:
    runs-on: ubuntu-latest

    steps:
      - uses: actions/checkout@v3

      - name: Setup Python
        uses: actions/setup-python@v4
        with:
          python-version: 3.10

      - name: Install dependencies
        run: pip install -r requirements.txt

      - name: Run MCP API Agent
        run: python -m qa_agent.api_agent_runner

🧰 Available Tools

📊 Real-World Use Cases

• Banking API automation (AML / KYC)

• Collateral management system testing

• Microservices regression testing

• AI-driven QA automation frameworks

⚠️ Troubleshooting

❌ Port conflict

netstat -ano | findstr :5000
taskkill /PID <pid> /F

❌ Module error

pip install -r requirements.txt

🚀 Future Enhancements

• OpenAI / LLM integration

• UI dashboard for test execution

• Kubernetes deployment

• Advanced embedding-based RAG

• Postman collection auto-import

👨‍💻 Summary

This project demonstrates:

✔ AI-powered API testing ✔ MCP-based tool orchestration ✔ RAG-enhanced validation ✔ Enterprise-grade QA automation architecture