All glossary terms
A Protocol Interoperability

Agent-to-Agent Protocol (A2A)

A2A is the open standard for agents discovering each other and delegating tasks across platforms. It works, it is stable, and it treats the remote agent as a black box. Which means A2A is the first protocol that lets you delegate work you cannot observe.

Definition

The Agent-to-Agent protocol (A2A) is an open standard defining how independent AI agents discover one another and delegate tasks across platform and organizational boundaries. Announced by Google in April 2025 and donated to the Linux Foundation that June, it reached version 1.0 in 2026. It standardises delegation while deliberately treating the remote agent as opaque.

What is the Agent-to-Agent protocol?

A2A is a standard for one agent handing work to another agent it did not build. It covers three things: how an agent advertises what it can do, how a task is structured and tracked, and how both travel over the wire. It is the horizontal counterpart to tool protocols.

The governance history matters because it explains why the protocol has traction. Google announced A2A in April 2025. That June, it donated the specification and SDKs to the Linux Foundation, under Apache 2.0. Founding partners included Amazon Web Services, Cisco, Microsoft, Salesforce, SAP and ServiceNow. That removed the single-vendor question early, and that's usually the thing that kills a proposed standard.

By its one-year mark, the Linux Foundation reported more than 150 supporting organizations. The protocol had native integration in the major cloud agent platforms. It also had production deployments across supply chain, financial services, insurance and IT operations. IBM's Agent Communication Protocol was folded into the same effort in August 2025, which further reduced fragmentation. The protocol reached version 1.0 in 2026, and version 1.0.1 in May 2026 added an extension mechanism.

This page is the specification-level companion to agent interoperability. That page covers the four layers of cross-platform cooperation and where each stands. This one covers what A2A actually specifies, what it deliberately leaves out, and the consequence of its central design decision. If you want the landscape rather than the protocol, start there.

How does A2A actually work?

An agent publishes an Agent Card at a well-known URL declaring its skills and how to reach it. A client agent reads that card, submits a task, and tracks it through an eight-state lifecycle until an artifact comes back or the task fails.

The mechanics are deliberately unexotic, built on JSON-RPC 2.0 over HTTP with gRPC and HTTP-plus-JSON bindings also available. That conventionality is a feature, it means existing gateways, authentication and network controls apply.

A flow showing a client agent fetching a remote agent's Agent Card from a well-known URL, then submitting a task. The task moves through the eight lifecycle states, with input required and auth required shown as branches that return control to the client. Long-running tasks are tracked by server-sent events streaming or webhook push notifications. The remote agent returns an artifact and a final state, and its internal steps are shown as a shaded box marked not visible to the client.

The Agent Card

A JSON document published at a well-known path, describing the agent's skills, the content types it accepts, its transport bindings and its security schemes. This is the discovery mechanism, and it is deliberately simple: fetch a URL, read what the agent says it does. Note the phrasing. The card is self-declared, which is covered from the trust side at agent identity.

The task lifecycle

Work is modelled as a task moving through defined states, which is what makes long-running and multi-turn delegation tractable rather than requiring a synchronous call to stay open.

The eight states in the A2A task lifecycle and what each means for the calling agent
State What it means What the client does
submitted The task has been accepted and not started Wait, or subscribe to updates
working The remote agent is executing Stream updates, or poll
input_required The agent needs something more before it can proceed Supply it. This is where multi-turn delegation happens
auth_required Additional authorization is needed Obtain it, which is the credential question this protocol does not answer
completed Finished, with an artifact returned Consume the artifact
failed The agent could not complete it Decide whether to retry. The protocol does not decide for you
canceled Stopped before completion Handle a partial result if one exists
rejected The agent declined the task outright Route elsewhere

Long-running tasks are tracked across connections using server-sent events for streaming or webhook push notifications for genuinely asynchronous work. Failures surface either synchronously as a JSON-RPC error object or asynchronously as a transition to a failed or rejected state. Retry logic and circuit breaking are deliberately left to the client, which is a reasonable protocol decision and means the resilience work is yours.

How is A2A different from MCP?

MCP connects an agent downward to tools and data. A2A connects agents sideways to each other. The commonly used shorthand is that MCP is USB-C for tool connectivity and A2A is HTTP for agent collaboration, and production systems use both rather than choosing.

The directional framing is accurate and it understates the more consequential difference, which is what each protocol lets you see.

When an agent calls a tool over MCP, the tool acts like a function. It takes arguments, returns a result, and the calling agent's own trace records the whole exchange. When an agent delegates a task over A2A, things work differently. The remote agent is not a function; it is another agent making its own decisions. You receive its conclusion, not its reasoning. Both protocols abstract something. MCP abstracts an operation. A2A abstracts an actor.

The combination in practice is straightforward: A2A routes a task to the right specialist agent, and MCP gives that agent the context and tools it needs to do the work. They sit at different layers and there is no meaningful adoption trade-off between them. For the MCP side see agent frameworks, where convergence on MCP as the tool layer changed where framework lock-in sits.

Why does the remote agent being opaque matter?

Because everything you built to govern your own agents stops at the boundary. You get a task state and an artifact. You do not get the remote agent's steps, its tool calls, its cost, or the authority it exercised, so four of your controls go dark at once.

This is the consequence least discussed in A2A material, and it is the one that determines whether cross-boundary delegation is something you can defend. The opacity is not a flaw in the protocol; it is the point of the protocol. Cooperation between agents built by different organizations requires exactly this abstraction. It also has a bill.

On the left, your estate showing four controls in place: a registry entry with an owner, per-step traces, evaluation scores, and an audit record with attribution. In the centre, the A2A boundary. On the right, the remote agent as a shaded black box returning only a task state and an artifact. Four crossed-out markers at the boundary show the registry entry, trace, evaluation and attribution failing to cross it.
  • Your inventory does not cover it. The remote agent is not in your agent registry and cannot be. You can register the fact that you delegate to it, which is worth doing, and you cannot record what it is or who changed it last.
  • Your traces stop at the boundary. Per-step tracing is what makes an agent failure investigable, and a delegated task returns a result rather than a span tree. A failure inside the remote agent is a status code to you. See AI agent observability.
  • Your evaluation cannot score the trajectory. Agent evaluation depends on judging the path as well as the outcome. Across an A2A boundary you can score the artifact and nothing else, so a remote agent that reached the right answer by an unacceptable route passes.
  • Your audit record loses the authority chain. The record can show that you delegated a task and what came back. It cannot show on whose authority the remote agent acted, because cross-organizational delegated authority is not standardised. See AI audit trail.

Traceability is an optional extension, not a core guarantee. Four official example extensions exist, one of which addresses traceability, and none addresses governance. That is the right architecture for a protocol and it tells you where the work sits: if you need a remote agent's trace, you negotiate it commercially and require the extension contractually. Treat cross-boundary observability as a procurement requirement rather than a protocol feature, because that is what it currently is.

The workable stance is to treat a delegated task the way you would treat a subcontracted service rather than an internal function call. Record the delegation, capture what you sent and what came back, define what an acceptable artifact looks like so you can evaluate the output even without the path, and accept that a portion of your evidence chain now depends on a counterparty rather than on your own instrumentation.

What does A2A not specify?

There are four things worth knowing before you plan around it. What a declared capability means. How agents should be coordinated. How governance travels across the boundary. And what words like customer or vendor mean to both sides. Each gap is deliberate, and each one lands on you.

Knowing the boundaries prevents the common planning error of assuming the protocol covers a layer it never claimed.

Four things the A2A protocol does not specify, and where each has to be handled instead
The gap What it means in practice
Capability semantics An Agent Card declares skills in its own words. Two vendors both claiming to summarise documents mean different things, so automated agent selection is not reliable. Discovery is solved; matching is not
Shared meaning The protocol governs communication, not knowledge. If two agents hold different definitions of "active customer" or "approved vendor," they will produce contradictory output while communicating perfectly. See data intelligence
Orchestration A2A makes delegation possible and says nothing about which agent should handle which step, in what order, or what happens on partial failure. See multi-agent orchestration
Governance Nothing in the core specification carries policy, risk classification, or an accountability chain across the boundary. The published extensions do not address it either

Two smaller technical gaps are worth flagging for anyone implementing. The specification is described in prose rather than formally, so some constraints are ambiguous, including whether a canceled task may be resumed. And practitioner guidance consistently notes the need for application-level workarounds around per-skill schemas, token downscoping and registry standardisation. Adjacent protocol work is filling some of this in, including an agent payments effort extending A2A into regulated transactional settings, so expect the gap list to shorten.

Should you adopt A2A now?

Yes if you genuinely need agents from different vendors or organizations to cooperate, and it is a poor reason to restructure anything internally. The specification is stable and the adoption is real. The question is whether you have the problem it solves.

The honest assessment splits into three cases, and most organizations are in the third.

  • Adopt it now if you are crossing a boundary. Multiple vendor agents that must cooperate, or delegation to a partner organization's agent. This is the problem A2A exists for, the alternative is a bespoke integration per counterparty, and the protocol is stable enough to build on.
  • Adopt it opportunistically if your platform already speaks it. The major cloud agent platforms have integrated A2A natively, so you may be able to expose or consume agents over it at low cost. Worth doing for the optionality even without an immediate need.
  • Do not restructure internal agents around it. Inside one estate you already have shared identity, shared tracing and shared governance, and A2A's central abstraction throws exactly those away. Using a cross-organizational protocol between two of your own agents buys interoperability you do not need and pays for it in the visibility you just built.

One planning note. Since cross-boundary observability, delegated authority and capability matching are all currently implementation problems rather than protocol features, the cost of an A2A integration is not the protocol work. It is the surrounding apparatus, and that is where the estimate usually goes wrong.

Frequently asked questions about the A2A protocol

If your question is not here, our team will answer it directly.


Talk to a Specialist →
What is the A2A protocol?

A standard for one agent handing work to another agent it did not build. It specifies three things: how an agent advertises what it can do, how a task is structured and tracked through its life, and how both travel over the wire. Google announced it in April 2025 and donated it to the Linux Foundation that June, which removed the single-vendor question that usually kills a proposed standard before it gets traction.

Who controls the A2A protocol?

The Linux Foundation, since June 2025, under Apache 2.0. Google developed it and donated the specification and SDKs, with founding partners including Amazon Web Services, Cisco, Microsoft, Salesforce, SAP and ServiceNow. IBM's Agent Communication Protocol was folded into the same effort in August 2025, reducing fragmentation further. That governance structure is the main reason to treat A2A as a durable standard rather than one vendor's proposal.

Is A2A production ready?

The specification is. A2A reached version 1.0 in 2026, with version 1.0.1 in May 2026 adding an extension mechanism, and the Linux Foundation reported more than 150 supporting organizations and native integration in the major cloud agent platforms at the one-year mark, alongside production deployments in supply chain, financial services, insurance and IT operations. The core data models and bindings are no longer a moving target; expect application-level work around the gaps.

What is an Agent Card?

A JSON document an agent publishes at a well-known path, describing its skills, the content types it accepts, its transport bindings and its security schemes. It is the discovery mechanism, and it is deliberately simple: fetch a URL and read what the agent says it does. The important word is "says." The card is self-declared, and no third party attests to what it claims, which makes discovery a solved problem and trust an unsolved one.

What are the A2A task states?

Eight: submitted, working, input required, auth required, completed, failed, canceled and rejected. Two of them do more work than they appear to. Input required is where multi-turn delegation happens, letting the remote agent ask for more before proceeding. Auth required surfaces the credential question the protocol does not itself answer. Long-running tasks are tracked across connections through server-sent events or webhook push notifications rather than by holding a call open.

What is the difference between A2A and MCP?

Direction, and more importantly visibility. MCP connects an agent downward to tools and data; A2A connects agents sideways to each other. The deeper difference is what each abstracts. A tool called over MCP is a function, and the calling agent's own trace records the whole exchange. An agent delegated to over A2A is another decision-maker, and you receive its conclusion rather than its reasoning. MCP abstracts an operation; A2A abstracts an actor.

Do you have to choose between A2A and MCP?

No, and framing it as a choice is a category error. They sit at different layers and production systems routinely use both: A2A routes a task to the right specialist agent, and MCP gives that agent the context and tools to do the work. There is no adoption trade-off. If a vendor presents them as competing options, that is a signal about the vendor rather than about the protocols.

Can you trace a task delegated over A2A?

Not by default, and this is the consequence least discussed in A2A material. You receive a task state and an artifact, not the remote agent's steps, tool calls, cost, or reasoning. Traceability exists as one of four official example extensions rather than as a core guarantee. If you need a remote agent's trace, negotiate it commercially and require the extension contractually. Cross-boundary observability is currently a procurement requirement rather than a protocol feature.

Does A2A handle orchestration?

No. It establishes that a remote agent can be discovered and given a task, which makes cooperation possible across boundaries. It says nothing about which agent should handle which step, in what order, what happens when one fails, or how partial results are reconciled. Those are orchestration decisions and they remain yours. The two arrive together in practice, because the first thing teams do after connecting agents across platforms is discover they now have to coordinate them.

Does A2A solve capability discovery?

It solves discovery and not matching. An Agent Card declares skills in the operator's own words, so two vendors both claiming to summarise documents may mean quite different things by it. You can find agents and read what they claim; automatically selecting the right one for a task is not reliable on that basis. In practice teams maintain a curated list of counterparties they have actually evaluated, which is discovery done by humans once rather than by machines continuously.

Should you use A2A between your own internal agents?

Generally not. Inside one estate you already have shared identity, shared tracing and shared governance, and A2A's central abstraction deliberately discards exactly those. Using a cross-organizational protocol between two of your own agents buys interoperability you do not need and pays for it in the visibility you just finished building. Use it where you are genuinely crossing a boundary: another vendor's agent, or another organization's.

What does an A2A integration actually cost?

Less in protocol work than teams expect and more in everything around it, which is where estimates go wrong. The specification is conventional, built on JSON-RPC over HTTP, so existing gateways, authentication and network controls apply. The cost sits in the parts the protocol leaves to you: retry and circuit-breaking logic, cross-boundary authority, capability matching against counterparties you have evaluated yourself, and whatever observability you can negotiate rather than assume.

Delegate, and still see it
What happens to your audit trail at the boundary?

CAMS orchestrates third-party agents over the Agent-to-Agent protocol while keeping the delegation inside your own governance: registered counterparties, scoped credentials, spend ceilings, and a recorded trace of what you sent and what came back.