Currency Agent - GCP Architecture & Communication Flow Visualizer

Refined GCP & Local Developer Architecture Flow

Interactive component hierarchy with customizable arrowheads, directional flows, and step-by-step animation controls.

Arrow Type:

Flow Visual:

Execution Sequence Step

Local Developer Machine

A2A Client

(test_client.py)

Local A2A Agent

(agent.py - :10000)

gcloud Auth Proxy

(Cloud Run Proxy - :8080)

Google Cloud Platform (GCP)

Vertex AI API

(Gemini 2.5 Flash)

Cloud Run

(Private mcp-server)

Step 1: Client Query Dispatch over A2A JSON-RPC

test_client.py → agent.py (:10000)

Interaction Overview

The local client script sends a structured JSON-RPC request containing the query "Convert 250 USD to EUR" to the local A2A agent process.

Pass Direction

Request Pass

Protocol Wire

A2A JSON-RPC

Source Component

A2A Client

Target Component

Local A2A Agent

Protocol Payload Console

GCP Cloud Currency Agent Trace Sandbox

Execute live queries to trace request packets across local processes, gcloud Auth Proxy, Vertex AI, and Cloud Run.

Test Parameters

Conversion Amount

Base Currency

Target Currency

GCP Communication Log Idle

// Select parameters and click "Execute GCP Cloud Agent Trace" to start path trace.

Agent Synthesized Output: ---

Architecture Communication & Protocol Matrix

Protocol specifications, endpoints, authentication headers, and wire format bindings across boundary lines.

Source Node	Target Node	Connector Label	Transport / Port	Authentication	Data Schema
A2A Client	Local A2A Agent	A2A JSON-RPC	HTTP POST :10000	None (Local loopback)	A2A Task JSON Envelope
Local A2A Agent	Vertex AI API	gcloud Credentials	HTTPS REST / gRPC	Google OAuth2 Access Token	Gemini Function Call Payload
Local A2A Agent	gcloud Auth Proxy	HTTP / localhost	HTTP POST :8080	Local Proxy Loopback	FastMCP JSON-RPC 2.0
gcloud Auth Proxy	Cloud Run (Private mcp-server)	Authenticated HTTPS	HTTPS :443 (GCP Ingress)	GCP IAM ID Token Bearer	Authenticated MCP SSE / JSON-RPC

Process Deployment Boundaries & Security Topography

Separation between local developer workstation processes and Google Cloud Platform managed services.

Local Developer Machine Boundary

A2A Client

test_client.py

Process

Local A2A Agent

agent.py (Port 10000)

Port :10000

gcloud Auth Proxy

Cloud Run Proxy (Port 8080)

Port :8080

Google Cloud Platform (GCP)

Vertex AI API

Gemini 2.5 Flash

Managed API

Cloud Run

Private mcp-server

Private Container

Architecture Code Files

test_client.py

agent.py

mcp-server/server.py

test_client.py Python 3.10+ / A2A Protocol