Apigee Tutorial: Agentic Portal Quickstart For MCP, A2A & REST APIs
1. Introduction: The Need for an Agentic Portal
In the traditional API landscape, developers browse portals to find REST or GraphQL APIs to integrate into their applications. However, the rise of Generative AI has introduced new types of interfaces—specifically Tools (for LLMs) and Agents.
The Agentic Portal Quickstart is an experimental open source project on GitHub to connect world of APIs with the AI world of ecosystem agents. Since agentic integration is all built on APIs, this actually fits together perfectly.
The repository provides a ready-made frontend that serves as a catalog for:
MCP (Model Context Protocol) Servers
A2A (Agent-to-Agent) Interfaces
Standard REST APIs
It leverages Google Cloud’s Apigee API Hub as the backend registry and Apigee as the AI Gateway, giving you a centralized way to manage and expose these “Agentic APIs” to consumers.
Some people believe that we will not need portals in the agentic era, however I suspect we will still want to document, show & build trust through “official docs” for both traditional APIs as well as agentic tools & agents themselves.
Does this conflict with agent registeries in Gemini Enterprise or Vertex AI? No, since we are focusing on the API aspects of AI integration, and AI & agentic integration is all about APIs, a great public portal is still an important way to communicate and market your organization’s APIs, tools & capabilities.
2. 
Key Features
This portal is not just a static documentation site; it is a functional registry with interactive capabilities.
Unified Catalog: Browse standard REST APIs alongside AI-specific interfaces like MCP and A2A Agents.
Interactive Documentation: Explore and test APIs using integrated OpenAPI rendering (powered by Scalar).
User Management: Built-in support for users to register and log in using their Google Accounts.
Self-Service Access: Users can create app subscriptions and generate credentials to access your APIs seamlessly.
Agent Integration Ready: Provides ready-to-use code snippets for integrating these tools into agents built with the Agent Development Kit (ADK).
Real-Time Updates (Early Feature): Updates made in your Apigee API Hub are pushed immediately to the portal using Eventarc.
3. 
Prerequisites
Before deploying, ensure you have the following set up in your Google Cloud environment:
Google Cloud Project: An active GCP project.
Apigee & API Hub: You must have Apigee (Evaluation, Pay-as-you-go, or Subscription) and Apigee API Hub provisioned.
Cloud Run: Ensure the Cloud Run API is enabled on your project.
4. 
Deployment Steps
You can deploy this portal as a single container to Google Cloud Run. The repository includes helper scripts to streamline the process.
Option A: Google Cloud Shell (Recommended)
The easiest way to deploy is by using the pre-configured tutorial in Cloud Shell.
Click here to Open in Google Cloud Shell
Option B: Manual Deployment 
If you prefer to run the commands manually in your local terminal or standard shell:
Step 
: Clone and Configure Environment
Clone the repository and set up your environment variables.
git clone https://github.com/api-integration-samples/agentic-portal-quickstart.git
cd agentic-portal-quickstart
# Copy the template environment file
cp 1.env.sh .env
# Edit .env with your specific project details (Project ID, Region, etc.)
nano .env
# Load the variables
source .env
Step 
: Create Initial Resources
Run the pre-creation script to set up necessary dependencies.
./2.pre-create.sh
Step 
: Deploy the Service
This script builds the container and deploys it to Cloud Run.
./3.deploy-service.sh
Step 
: Post-Deployment Configuration
Run the final script to configure the portal after deployment.
./4.post-create.sh
5. 
Conclusion
The Agentic Portal Quickstart is an excellent starting point for organizations looking to modernize their developer experience for the AI age. By deploying this portal, you bridge the gap between traditional API management and the emerging world of autonomous agents, providing a single pane of glass for both human developers and AI consumers.
Want to see it in action first? Check out the live test deployment at portal.agenticplatform.dev.
Just reach out here in the comments or via message if you would like to dive deeper into this topic!
