Inspiration

I am interested in Google Cloud and AI Technologies hence I want to learn Google Agent Development Kit with some use case to experiment features like Tools Integration and also the associated Google ADK Java version

What it does

It enables users to search for hotels and accomodation with AI assistance leveraging various Agent Tools : MCP (ModelContextProtocol) Database Toolbox, MCP AirBnB external service , basic Weather Report Function Calling Agent

How we built it

Local version for Development , Test and Debugging

Docker Containers and Orchestration with Docker-Compose to facilitate the application Orchestration and Launch

Google Cloud Technologies : Cloud Shell, Cloud Run , Cloud Container Registry , Cloud Logging for serverless development and Deployment

Challenges we ran into

SSE access to MCP external service from ADK Java Application in Docker version

Fallback Strategy of the Agent in case of unavailability of some mcp services

Accomplishments that we're proud of

ADK Java Working Application with User Interface available locally, with Docker (Compose) containers and accessible as Google Cloud Run App Url

Integration to the Agent of MCP servers (Toolbox for Databases, AirBnB) to wrap access to external resources and services

Illustration of using Function Calling as Tool to extend Agent service

What we learned

Google ADK Agentic solution in Java

What's next for MyAgent

Finalization of Google Cloud other Features (Cloud SQL,...)

Completion of the Hotels Database

Built With

  • docker
  • goocle-cloud-logging
  • google-adk
  • google-cloud-container-registry
  • google-cloud-run
  • google-cloud-shell
  • google-gemini-llm
  • java
  • mcp
  • mcp-server-airbnb
  • mcp-toolbox
  • sse
Share this project:

Updates