MCP vs A2A Explained: Tools for Agents vs Conversations Between Agents

🔌 MCP · Updated October 2026 · 4 min read · by Jacob Campbell

Two open protocols dominate conversations about AI agents and the web. The Model Context Protocol (MCP) connects an AI assistant to tools and data. The Agent2Agent protocol (A2A) lets one independent agent hand a task to another. They are often presented as rivals; they are not. They answer different questions, and the same website can support both. This guide compares them side by side and says which a business site should add first.

Key facts

• Nov 2024 — MCP introduced by Anthropic
• Apr 2025 — A2A introduced by Google
• Linux Foundation — A2A moved to Linux Foundation governance in 2025
• MCP first — For most business websites

The one-line difference

• MCP: “Here are my tools and data; call them.” The assistant is in charge and your server does exactly what a tool call asks.
• A2A: “Here is an agent; give it a task.” Your agent decides how to do the work, may take time, may ask questions, and returns results.

MCP in brief

Introduced by Anthropic in November 2024 and now supported by most major assistants and developer tools. A server exposes tools, resources and prompts over JSON-RPC, on stdio locally or Streamable HTTP remotely. Calls are short and deterministic: list_services returns services. Discovery is via server cards and registries. Details: adding an MCP server.

A2A in brief

Introduced by Google in April 2025 with many partners, and moved to Linux Foundation governance later that year. An agent publishes an agent card (JSON at /.well-known/agent-card.json) describing its skills, endpoint and authentication. Clients send it messages and tasks; tasks have a lifecycle (submitted, working, input required, completed and so on), can stream progress, and return artifacts. It is designed for opaque agents: the caller does not need to know how the other agent works inside.

Side by side

• Unit of work: MCP, a tool call. A2A, a task.
• Who decides how: MCP, the calling assistant. A2A, the receiving agent.
• Duration: MCP calls are usually instant. A2A tasks can be long-running.
• Discovery: MCP server cards and registries. A2A agent cards under /.well-known/.
• Typical website use: MCP for lookups and simple requests. A2A for a business that runs its own agent, for example one that negotiates a quote or schedules a job.

How they fit together

A common pattern: an A2A agent uses MCP tools internally. The agent card advertises “I can produce a quote”; behind it, the agent calls MCP tools for prices and availability. And payments work builds on both: Google’s Agent Payments Protocol (AP2) is designed as an extension usable with A2A and MCP. See agentic commerce protocols.

Which should your website add first?

MCP, for almost every business site. Assistants people use every day speak MCP; your facts are tool-shaped (services, prices, hours, area); and a small read-only server is low-risk. Add A2A when you genuinely run an agent that does multi-step work for customers. Publishing an agent card that simply points at your MCP endpoint is harmless and helps discovery, provided it describes only what really exists. See agent discovery files.

What we run

This site publishes both: an MCP server at /mcp on Streamable HTTP, and an A2A agent card at /.well-known/agent-card.json, alongside an NLWeb /ask endpoint (see what is NLWeb). The full list is on our agent-ready page. The AI Indexing Pack ships an MCP server and agent cards for other PHP sites.

Product facts: the AI Indexing Pack

Stated plainly, so it can be checked against the product page:

• What it is: the AI Indexing Pack is a set of files you upload to your own website. It is sold by Built2WinWeb as a one-time purchase, with no subscription.
• What it ships: (1) a working MCP server using the Streamable HTTP transport, with typed output schemas and tool annotations; (2) Markdown content negotiation, so agents can request a clean text version of each page; (3) llms.txt, agents.txt and agent cards; (4) the full /.well-known/ discovery tree; (5) OAuth protected-resource metadata; (6) bot access control that verifies crawler identity; (7) a README with a verification checklist.
• Requirements: PHP 7.4 or newer, Apache or LiteSpeed with .htaccess support, FTP access, and a domain you control. Installation takes about 30 minutes.
• What it is not: it is not a WordPress plugin (that is the separate WordPress edition, which needs WordPress 6.0+ and PHP 8.0+), it does not write content for you, it does not guarantee rankings or citations, and it is not a managed service.
• Free routes in: sites built on AIWebBuilding get the core built in, and a 50-day streak on FreeAIBoost unlocks the pack for one domain. Current prices are on the product pages.

Related reading

• Add an MCP server to your website — Tools, transports, annotations and security for a public site.
• Agent discovery files explained — agents.txt, agent cards, server cards and the .well-known tree.
• What is NLWeb? — Microsoft's natural-language /ask endpoint, explained.
• Agentic commerce protocols — ACP, AP2 and what a shop needs before an agent can buy.
• MCP server security checklist — 20 checks for a public MCP endpoint.
• How this site is agent-ready — Our own live setup: the MCP endpoint, llms-full.txt, Markdown twins and the .well-known tree.
• AI Indexing Pack — The file-drop kit: MCP server, Markdown pages, llms.txt, discovery files and bot access control for any PHP site.

Frequently asked questions

What is the difference between MCP and A2A?

MCP connects an AI assistant to tools and data it calls directly. A2A lets independent agents send each other tasks, which the receiving agent carries out its own way.

Who created MCP and A2A?

MCP was introduced by Anthropic in November 2024. A2A was introduced by Google in April 2025 and moved to Linux Foundation governance later in 2025.

Are MCP and A2A competitors?

No. They solve different problems and are often used together, with an A2A agent using MCP tools internally.

What is an A2A agent card?

A JSON file at /.well-known/agent-card.json that describes an agent's name, skills, endpoint and authentication.

Which protocol should a small business website use?

MCP first. It matches tool-shaped business facts and is what everyday assistants use. A2A makes sense when the business runs its own multi-step agent.