How zeeeepa/mcp-aas works
Main Components
• Frontend: Built with React. Responsible for presenting the user interface and integrating with AWS Amplify for Cognito-based authentication.
• Backend: A Node.js/Express service that handles core API functionalities (user management, tool launching, routing) including authentication logic with AWS Cognito.
• Infrastructure: Contains AWS CDK (and Terraform) code for deploying resources such as AWS Cognito, AWS ECS, Lambda functions, and other cloud services.
• Shared Libraries/Utilities: Found in the “shared” directory and reused by both frontend and backend.
• Specialized Python Service: The “mcp-tool-crawler-py” module, which appears to handle crawling and source management tasks. This service involves CLI scripts, lambda functions, and integrates with AWS step-functions.
• Testing/CI Components: End-to-end tests (in folder “e2e”), unit tests (in __tests__ directories), and GitHub Actions workflows (.github/workflows) for CI/CD processes.
Relationships & Data Flow
• Users (clients) interact with the Frontend (React app) via browsers.
• The Frontend communicates with the Backend via RESTful APIs. This layer could include sessions, JWT tokens, and WebSocket connections to tools.
• The Backend directly interacts with AWS Cognito for authentication (user registration, authentication, and profile management).
• Infrastructure components (AWS CDK/Terraform scripts) are responsible for provisioning services like Cognito, ECS containers (hosting both the frontend and backend docker images), Lambda functions, and event handling services.
• The Python-based “mcp-tool-crawler-py” service likely operates as an independent microservice, which might be triggered by events (possibly managed through AWS Step Functions) to crawl sources and manage tool packaging.
• Shared utility modules ensure consistent behavior across different services.
Architectural Patterns & Design Principles
• Microservice components: The separation between the frontend, backend, and the independent Python crawler service illustrates a microservice-oriented approach.
• Separation of Concerns: Frontend logic remains within the “frontend” folder, backend API handling in “backend”, and infrastructure orchestration is managed separately.
• Infrastructure as Code (IaC): Using AWS CDK (and Terraform) for automated deployment ensures reproducible cloud configurations.
• Testing: A robust test suite including unit tests (under __tests__) and end-to-end tests (in “e2e”) is in place.
Technologies
• Frontend: React, AWS Amplify
• Backend: Node.js, Express, AWS SDK for integrating with Cognito
• Infrastructure: AWS CDK, AWS ECS, AWS Cognito, Terraform modules
• Python Service: Python scripts, AWS Lambda integrations, AWS Step Functions (via the provided JSON definition)
• Containerization: Docker (both for frontend and backend) and Docker Compose for local testing.
• CI/CD: GitHub Actions to manage build and deployment pipelines.