🤖 Build an Autonomous Transit Agent with Google ADK & MCP

Executive Summary

Today, we are moving beyond standard chatbots. In this hands-on lab, you will build a deterministic, multi-reasoning Agentic AI system using Google's Agent Development Kit (ADK) and the Model Context Protocol (MCP).

Instead of an AI that just generates text, you will build an orchestrator. Our Transit Agent will dynamically fetch live data, read business policies, and execute transactional workflows completely on its own. Finally, we will serve it through a beautiful web UI!

The Tech Stack:

- Google ADK: The "Brain" - Orchestrates the ReAct (Reason + Act) loop using Gemini.

- FastMCP: The "Muscle" - Exposes your backend APIs and logic to the AI.

- Docker: For reliable, instant deployment.

☁️ Google Cloud Setup (Do This First!)

Before we look at the code, we need a workspace. Follow these steps to set up your Google Cloud environment:

Step 1: Create a Google Cloud Project

Open the Google Cloud Console.

Click the Project Selector dropdown at the top of the page.

Click New Project.

Name your project (e.g., gdg-mumbai-agent) and click Create.

Once created, make sure your new project is selected in the top dropdown.

Step 2: Enable billing and redeem the credits

Make sure that billing is enabled for your Cloud project. Learn how to check if billing is enabled on a project.

Redeem Credits following the setps mentioned in codelab.

Step 3: Activate Cloud Shell

We will use Google Cloud Shell as our terminal so we don't have to install anything locally.

Look at the top right corner of the Google Cloud Console.

Click the Activate Cloud Shell icon (it looks like this: >_).

A terminal will open at the bottom of your screen. Click Continue if prompted.

Once connected to Cloud Shell, you check that you're already authenticated and that the project is set to your project ID using the following command:

gcloud auth list

Run the following command in Cloud Shell to confirm that the gcloud command knows about your project.

gcloud config list project

If your project is not set, use the following command to set it:

gcloud config set project <YOUR_PROJECT_ID>

Step 4: Enable the Required APIs

To use Google's generative models, we need to turn on the APIs for this project.

In the Google Cloud Console search bar at the top, type "Vertex AI API" and select it.

Click Enable.

Next, search for "Generative Language API" (used for Gemini API keys) and click Enable.

Alertnatively, run this command in cloud shell:

gcloud services enable aiplatform.googleapis.com generativelanguage.googleapis.com

(Note: If prompted to authorize Cloud Shell, click Authorize).

🚀 Quick Start: Google Cloud Shell

Still in your Cloud Shell open, simply paste this command into your terminal:

git clone https://github.com/ruby-verma/transit-agent.git && cd transit-agent 

🔑 Set Up Your API Key

You will need a Gemini API key to give your agent a brain.

Get your API key from Google AI Studio.

Once you have your key, choose one of the following methods to add it to your project:

Option A: The Quick CLI Method (Recommended)

Run this single command in your Cloud Shell terminal, making sure to replace the placeholder with your actual copied key:

In your Cloud Shell terminal, copy the example environment file:

echo 'GOOGLE_API_KEY="YOUR_API_KEY"' > .env

Option B: The Visual Editor Method

Copy the template file:

cp .env.example .env

Use the Cloud Shell Editor's file explorer on the left side of your screen.

Open the .env file, paste your key inside, and save (Ctrl+S).

📂 Repository

... View full README on GitHub