Finda2aFinda2a
Back Home/Server List
S

Semantic Kernel Agent with A2A Protocol

Featured Service
Default
28 views

Introduction

This sample demonstrates how to implement a travel agent built on Semantic Kernel and exposed through the A2A protocol. It showcases: Multi-turn interactions: The agent may request clarifications Streaming responses: Returns incremental statuses Conversational memory: Maintains context (by leveraging Semantic Kernel’s ChatHistory) Push notifications: Uses webhook-based notifications for asynchronous updates External plugins (SK Agents & Frankfurter API): Illustrates how an Semantic Kernel Agents are used as plugins, along with APIs, that can be called to generate travel plans and fetch exchange rates

#LLM
#UV
#Python 3.10
#Frankfurter API key
#SK setup

Features

  • Multi-turn Interactions
    The agent can request clarification from users to gather missing information before completing a task.

  • Streaming via Incremental State Updates
    Instead of one-time responses, the agent provides progressive feedback using TaskStepUpdates.

  • Conversation Memory
    Uses Semantic Kernel’s ChatHistory to track user-agent conversations and maintain context.

  • Push Notifications
    Supports out-of-band notifications via webhook to inform the A2A server of updates asynchronously.

  • Plugin Integration
    Extends capabilities using additional agents and APIs (e.g., exchange rates, activity planning).