Agentic AI Communication Protocols logopeterindia.net
Knowledge Hub

Agentic AI Communication Protocols: MCP, A2A, Agent Client Protocol & More

How AI agents talk to tools, to each other, to editors and to users — MCP, A2A, the two protocols called ACP, AG-UI and the newer commerce protocols — with specifications, SDKs and guidance on combining them.

59 topics•8 categories•Updated October 2026

How the Protocols Fit Together

6 entries

Most agent protocols solve one link in the chain. MCP connects an agent or model to tools and data. A2A connects an agent to other agents so they can discover each other, delegate tasks and exchange results. Agent Client Protocol connects a code editor to a coding agent. AG-UI connects an agent to a user-facing application. Commerce and payment protocols such as AP2, UCP, the Agentic Commerce Protocol and x402 cover buying and paying. They are complementary rather than competing: a typical system receives a request through an API, uses A2A to delegate between agents, MCP for tool calls, and human-in-the-loop approval before a consequential action. Beware the name clash: two different protocols are called “ACP”.

01AI Agents (02): Agentic Communication Protocols (0s & 1s, Medium, May 2026)The article that prompted this page: step-by-step flows for MCP, A2A and ACP, and how they combine in one workflow. 02Agentic AI Foundation (AAIF)Linux Foundation home for open agentic standards; MCP was contributed to it. 03Linux Foundation announces the Agentic AI FoundationDecember 2025 announcement of AAIF and its founding projects. 04A2A and MCPThe A2A project's own explanation of how it complements MCP. 05JSON-RPC 2.0 specificationThe message format underneath MCP, A2A and Agent Client Protocol. 06AGENTS.mdSimple instruction file convention for coding agents, also an AAIF project.

Model Context Protocol (MCP): Agents to Tools

11 entries

Open standard for connecting models and agents to tools, data and prompts.

07Model Context ProtocolOfficial documentation, introduction and getting started guides. 08MCP specificationThe current specification, including tool discovery with tools/list, schemas and authorization. 09Introducing the Model Context Protocol (Anthropic)The November 2024 announcement. 10MCP specification repositorySpecification and documentation source. 11MCP reference serversReference and community server implementations. 12MCP Python SDKOfficial Python SDK. 13MCP TypeScript SDKOfficial TypeScript SDK. 14MCP InspectorVisual tool for testing and debugging MCP servers. 15FastMCPPythonic framework for building MCP servers and clients. 16MCP in the OpenAI Agents SDKUsing MCP servers from a popular agent framework. 17Top MCP ServersCurated MCP servers on this site.

Agent2Agent (A2A): Agents to Agents

7 entries

Discovery, delegation and collaboration between independent agents.

18A2A ProtocolOfficial site: concepts, tutorials and the specification. 19A2A specificationAgent Cards, tasks, messages and streaming. 20A2A on GitHubReference repository; an open source project under the Linux Foundation, contributed by Google. 21A2A Python SDKOfficial Python SDK. 22A2A JavaScript SDKOfficial JavaScript SDK. 23A2A samplesExample agents and clients. 24Announcing the Agent2Agent Protocol (Google)The April 2025 launch post.

Two Protocols Named ACP

5 entries

Do not confuse them: one links editors to coding agents, the other has merged into A2A.

25Agent Client Protocol (ACP)Standard for connecting code editors to coding agents, started by Zed; uses JSON-RPC and lets the editor approve tool calls such as file writes. 26Agent Client Protocol on GitHubSpecification and SDKs. 27Zed: Agent Client ProtocolZed's overview and list of supported agents. 28Agent Communication Protocol (IBM / BeeAI)A separate REST-based agent-to-agent protocol from IBM Research's BeeAI project; it has since joined A2A. 29Agent Communication Protocol repository (archived)Archived; the README says ACP is now part of A2A under the Linux Foundation and links a migration guide.

Agent-to-User Interfaces

3 entries

Streaming agent state to front-end applications.

30AG-UI documentationAgent-User Interaction Protocol: event-based, bidirectional agent-to-frontend communication. 31AG-UI on GitHubProtocol, SDKs and integrations. 32NLWebMicrosoft-originated project for natural-language interfaces to websites that also exposes an MCP server.

Open Agent Networks & Gateways

6 entries

Discovery, identity and routing across the wider internet of agents.

33Agent Network Protocol (ANP)Open protocol for agent identity and communication across the internet. 34Agent Network Protocol on GitHubSpecification and reference code. 35AGNTCYOpen infrastructure for an “Internet of Agents”: discovery, identity and messaging. 36AGNTCY on GitHubSource repositories. 37agentgatewayProxy for agent traffic that understands MCP and A2A. 38ERC-8004: Trustless AgentsEthereum proposal for agent identity, reputation and validation registries; check its status before relying on it.

Agentic Commerce & Payment Protocols

9 entries

How agents buy things and pay.

39Agent Payments Protocol (AP2)Google-led protocol for authorising and auditing agent-initiated payments. 40AP2 on GitHubSpecification and samples. 41Universal Commerce Protocol (UCP)Open standard for commerce between agents, businesses and payment providers. 42UCP on GitHubSpecification repositories. 43Agentic Commerce ProtocolOpen standard for agent-assisted checkout from Stripe and OpenAI; not the same as the other two ACPs. 44Agentic Commerce Protocol on GitHubSpecification and schemas. 45x402HTTP 402-based payment protocol for machine-to-machine payments. 46x402 on GitHubReference implementation, now under the x402 Foundation. 47A2A x402 extensionExtension adding x402 payments to A2A.

Related Pages

12 entries

Neighbouring pages on this site.

48Agentic AIAgentic AI hub. 49Top Agentic AI FrameworksFrameworks that implement these protocols. 50AI Agent FrameworksAgent frameworks. 51AI Agent GatewaysGateways for agent traffic. 52AI Agent Meshes & InteroperabilityMeshes and interoperability. 53AI Agent Security SolutionsSecuring agents and tools. 54Guardrails for Agentic AIGuardrails. 55AI Agent GovernanceGovernance. 56AgenticOpsOperating agents. 57AI Agent ObservabilityObservability. 58MCP Design PatternsSix ways to structure MCP servers. 59MCP Servers for AI DevelopersTen MCP servers for AI coding agents.

Frequently Asked Questions

What are agentic AI communication protocols?

They are open standards that define how agents exchange messages with tools, other agents, editors, user interfaces and payment systems, so components from different vendors can work together without custom integrations. Examples are MCP, A2A, Agent Client Protocol, AG-UI and AP2.

What is the difference between MCP and A2A?

MCP connects an agent or model to external tools and data through a discoverable, schema-described interface. A2A connects agents to other agents so they can find each other, delegate tasks and return results. They are designed to work together: MCP for tool use and A2A for collaboration.

Which protocol is ACP?

Two different protocols share the name. Agent Client Protocol, started by Zed, links code editors to coding agents. The Agent Communication Protocol from IBM's BeeAI project was an agent-to-agent protocol that has since merged into A2A. A third, the Agentic Commerce Protocol, covers checkout. Always check which one a source means.

Who governs these protocols?

The Linux Foundation hosts A2A, and the Agentic AI Foundation, also under the Linux Foundation, hosts MCP. Other protocols have their own owners, so check each project's site for current governance.

How current is this page?

Links were checked in October 2026. Agent protocols are changing quickly: several projects have moved foundations, merged or been renamed, so confirm versions and governance on each project's own site before adopting. Sites that block automated checks are kept and may need a manual click.