langgraph-a2a-client 0.1.0

A2A Protocol Client Tool for LangGraph

LangGraph A2A Client

A2A (Agent-to-Agent) Protocol Client Tool for LangGraph.

Overview

This library provides functionality to discover and communicate with A2A-compliant agents in LangGraph applications.

Key Features

• Agent discovery through agent cards from multiple URLs
• Message sending to specific A2A agents
• LangChain tool integration for easy use in LangGraph workflows

Installation

pip install langgraph-a2a-client

Or with uv:

uv add langgraph-a2a-client

Usage

from langgraph_a2a_client import A2AClientToolProvider

# Initialize the A2A client
a2a_client = A2AClientToolProvider(
    known_agent_urls=["https://example.com/agent"],
    timeout=300,
    webhook_url="https://your-webhook.com/notify",
    webhook_token="your-webhook-token"
)

# Get the tools for use in LangGraph
tools = a2a_client.tools

# Use in your LangGraph application
# ...

API

A2AClientToolProvider

Main class that provides A2A client functionality.

Parameters

• known_agent_urls (list[str] | None): List of A2A agent URLs to discover initially
• timeout (int): Timeout for HTTP operations in seconds (default: 300)
• webhook_url (str | None): Optional webhook URL for push notifications
• webhook_token (str | None): Optional authentication token for webhook notifications

Tools

The provider exposes three tools:

1. a2a_discover_agent: Discover an A2A agent and return its agent card
2. a2a_list_discovered_agents: List all discovered A2A agents and their capabilities
3. a2a_send_message: Send a message to a specific A2A agent

Examples

Basic Usage

uv run examples/basic_usage.py

Supervisor Agent Example

export OPENAI_API_KEY="your-openai-api"
uv run --extra examples examples/supervisor_agent.py

License

MIT