A2A 規範未言盡的實踐細節與深度指南

依據 A2A v1.0 規範撰寫的欄位級智慧體發現實踐指南。每篇指南均配合實際檢測與驗證工具,助您快速落地。

部分文章提供繁體中文譯本,未翻譯文章以英文展示。

精選深度指南

全部技術指南與解析

架構

智慧體架構大一統:Linux 基金會 AAIF 下的 A2A 與 MCP 協同實戰

詳解 Linux 基金會 Agentic AI Foundation (AAIF) 如何將 MCP 與 A2A 整合為生產級雙層架構,實現內部工具呼叫與跨框架外部協同的深度統一。

10 分鐘閱讀
生態

A2A x402 商業化擴充:如何為 AI 智慧體配置按次即時付費呼叫

解析 A2A x402 協定擴充如何將 HTTP 402 Payment Required 引入 Agent Card,讓自主智慧體支援定價宣告、計量計費與跨智慧體即時清結算。

9 分鐘閱讀
Comparison

A2A vs MCP: two protocols solving two different problems

Compare A2A vs MCP: what each protocol connects, where they overlap, and why production AI agent systems often use both for tools and agent collaboration.

9 min read
How-to

How to create an agent.json file for A2A v1.0

Learn how to create an agent.json file for A2A v1.0, including required Agent Card fields, the well-known URL, response headers, and common mistakes.

11 min read
Integration

Publishing an Agent Card from ADK, LangGraph, CrewAI or Semantic Kernel

Learn how to publish an A2A Agent Card from Google ADK, LangGraph, CrewAI, or Semantic Kernel by mapping framework concepts to standard discovery fields.

10 min read
Troubleshooting

Agent Card validation errors explained, one by one

Diagnose every A2A v1.0 Agent Card validation error and warning, understand what triggers it, and apply the exact fix in validator check order.

12 min read
Protocol

What A2A v1.0 changed in the Agent Card

Learn how A2A v1.0 changes Agent Cards, from supportedInterfaces and required skill tags to removed fields, with practical v0.x migration examples.

9 min read