A2A MCP Client Server

You can connect your MCP client to an Agent-to-Agent (A2A) companion, enabling seamless messaging, task management, and streaming responses through the Model Context Protocol. This server acts as a bridge so your LLMs can interact with A2A agents in a structured, MCP-friendly way.

How to use

You will run the MCP server locally or in a container and connect your MCP client to it. Use the available tools to send tasks, monitor progress, stream updates, and fetch agent information. Typical patterns involve sending a task, polling or streaming updates, and querying agent details to tailor requests to the agent’s capabilities.

Key usage patterns include: sending a task to an A2A agent to perform a computation or generate content; subscribing to updates for real-time feedback; retrieving the current status of tasks; and querying agent information to adapt your prompts to the agent’s capabilities.

How to install

Prerequisites you need before installation:

• Node.js (stable LTS release) and npm are required to install and run the MCP server.

Install options you can choose from.

# Install globally
npm install -g a2a-client-mcp-server

# Or run directly with npx
npx a2a-client-mcp-server

Configuration and usage examples

Set the A2A endpoint URL to point to the A2A agent you want to connect to. The default is http://localhost:41241 if you do not provide a value.

# Example configuration for Claude Desktop integration (NPX approach)

{
  "mcpServers": {
    "a2a_client": {
      "command": "npx",
      "args": ["-y", "a2a-client-mcp-server"],
      "env": {
        "A2A_ENDPOINT_URL": "http://localhost:41241"
      }
    }
  }
}

If you prefer running via Docker, build and run as shown below to connect with the same endpoint:

docker build -t a2a-client-mcp-server .

# Run with environment variable exposed to the container
docker run -i --rm -e A2A_ENDPOINT_URL=http://localhost:41241 a2a-client-mcp-server

Notes on connections and environment

This server can be started using a local runtime (stdio) approach. The command to start is issued as a single runtime call with arguments that include the package and any required flags. The environment variable A2A_ENDPOINT_URL controls which A2A agent is targeted by the MCP server.

Development

To set up a development environment, install dependencies and start a watcher that rebuilds as you code.

# Install dependencies
npm install

# Build the project
npm run build

# Run in development mode
npm run watch

Available tools

a2a_send_task

Send a task to an A2A agent with an optional taskId to track progress.

a2a_get_task

Retrieve the current state of a previously submitted task using its taskId.

a2a_cancel_task

Cancel a running task identified by its taskId.

a2a_send_task_subscribe

Send a task and subscribe to its updates in streaming mode, with an optional maxUpdates limit.

a2a_agent_info

Query information about the connected A2A agent, including capabilities and metadata.