Introducing MuleSoft A2A Support

We are excited to announce the release of MuleSoft Agent2Agent (A2A) Protocol Support (Beta) across MuleSoft’s Anypoint Platform. These powerful new capabilities build on Google’s open-source A2A protocol and enable AI agents to securely communicate, collaborate, and take coordinated action across complex enterprise environments.

In conjunction with our recent MCP support release and governance for AI agent interactions release, A2A support will help our customers further accelerate their AI transformation by providing the protocols, tools, and governance needed to build secure, scalable, multi-agent systems. By simplifying how teams can orchestrate AI agents across the enterprise, MuleSoft is helping lay the foundation for scalable, multi-agent ecosystems.

AI agents are quickly gaining traction across enterprises, helping teams eliminate manual, repetitive tasks and improve efficiency. But as organizations scale from one-off use cases to multi-agent systems, they face a fundamental challenge: how do you enable agents to interact, share knowledge, and delegate tasks securely across different systems?

That’s where Agent2Agent (A2A) protocol comes in.

A2A is an emerging open standard that defines how agents can discover, invoke, and coordinate with other agents in a secure and governed way. And now with MuleSoft A2A Support, customers can:

• Build multi-agent workflows: Bring the A2A protocol to any AI agent, like Agentforce or an internal agent, in just a few clicks using the MuleSoft A2A Connector 
• Govern & control agentic communication: Secure agent access, detect and filter PII, modify requests, and get centralized oversight of agent interactions
• Simplify agent discoverability: Use a centralized registry of agent-ready services to accelerate agent development (H2 '25)

Together, these capabilities make it easier to scale multi-agent use cases across departments, domains, and environments.

Designed with developer productivity and enterprise readiness in mind, MuleSoft A2A Support helps teams move fast, without compromising security or control.

MuleSoft A2A Connector (Beta): The MuleSoft A2A Connector makes it easy for developers to expose any existing agent as an A2A-compliant server or client. In just a few steps, teams can define agent capabilities, register them with the A2A registry, and enable secure communication between agents. The connector simplifies the complexity of implementing A2A standards by abstracting boilerplate code and providing a prebuilt listener and client configuration. Developers can focus on business logic, not protocol wiring, to get agents up and running faster.

A2A Policies for Anypoint Flex Gateway (GA): While the connector enables agents to talk to one another, Anypoint Flex Gateway ensures they do so securely and with full visibility. With the new out-of-the-box A2A policies, Flex Gateway acts as a secure and governed gateway layer for agentic traffic. Organizations can apply consistent policies, control access to agent resources, and monitor interactions across all deployed agents, just as they would for APIs. Learn more

With MuleSoft A2A Support, organizations can unlock entirely new levels of automation, coordination, and intelligence, such as:

Cross-functional agent collaboration for task orchestration

The MuleSoft A2A Connector allows developers to easily expose agents as A2A-compliant servers and clients, enabling seamless communication between task-specific agents. This makes it easy to break down complex business workflows into modular agent interactions.

An example of AI agents in action: An order processing agent needs to coordinate with both a pricing agent and an inventory agent before confirming an order. Using the MuleSoft A2A Connector, each agent exposes specific capabilities — like "calculate discount" or "check stock level" — as discoverable A2A endpoints. When a customer places an order, the order processing agent automatically invokes these actions from the other agents to retrieve pricing and inventory data, enabling real-time, multi-agent collaboration to finalize the purchase.

Domain-based security and controlled agent discovery

As organizations scale their use of AI agents, it becomes critical to ensure that each agent only accesses the resources it’s authorized to use. A2A policies in Flex Gateway provide governance over which agents can discover and interact with each other — enforcing the principle of least privilege.

An example of AI agents in action: A front-line customer support agent handles a return request and needs to coordinate with a return processing agent. While the support agent has access to customer information, the return agent is limited to logistics workflows. Flex Gateway’s A2A policies ensure that only authorized data is shared between the two agents, while allowing seamless task delegation. The support agent collects any additional inputs from the customer and passes it on to the return agent via a secured A2A interaction ensuring data control across domains.

With A2A and MCP, MuleSoft is further evolving from system-to-system integrations to agent-to-system and agent-to-agent interactions. It’s all part of our broader mission to help every organization build the infrastructure needed for a scalable, secure, and composable AI future.

MuleSoft A2A Support is currently in beta in Anypoint Platform. Additional platform-wide capabilities — including observability tooling and expanded agent registration — will roll out over the coming months.

As the future of work shifts toward intelligent, autonomous systems, MuleSoft remains committed to providing our customers with the latest in AI and agentic innovation to help them embrace the future with agility and confidence. 

Learn more about A2A Support and how to get started