feat: Add native A2A (Agent-to-Agent) protocol support

[ankitlade12]

Summary

This PR adds native support for Google's Agent-to-Agent (A2A) protocol to LangGraph, enabling LangGraph agents to communicate with agents from other frameworks.

Motivation

The A2A protocol is emerging as a standard for agent interoperability in 2025-2026. This feature allows LangGraph agents to:

• Be discovered by other agents via Agent Cards
• Receive and process standardized task requests
• Respond using A2A-compliant message formats

Changes

New Module: langgraph/a2a/

File	Description
capabilities.py	A2ACapabilities class for defining agent skills
card.py	AgentCard and AgentEndpoint for discovery
message.py	A2AMessage, A2ARequest, A2AResponse types
protocol.py	A2AProtocolHandler for wrapping LangGraph agents

Tests

• Added 46 comprehensive test cases in tests/test_a2a.py
• All tests passing
• All linting checks passing (ruff, mypy)

Example Usage

from langgraph.graph import StateGraph
from langgraph.a2a import A2ACapabilities, A2AProtocolHandler

capabilities = A2ACapabilities(
    name="research-agent",
    description="Performs web research",
    skills=("search", "summarize"),
)

graph = StateGraph(MyState)
# ... build graph
compiled = graph.compile()

handler = A2AProtocolHandler(graph=compiled, capabilities=capabilities)
card = handler.get_agent_card()  # For discovery
response = handler.handle_request(request)  # Handle A2A requests

[vercel]

@ankitlade12 is attempting to deploy a commit to the LangChain Team on Vercel.

A member of the Team first needs to authorize it.