Checked other resources

• This is a feature request, not a bug report or usage question.
• I added a clear and descriptive title that summarizes the feature request.
• I used the GitHub search to find a similar feature request and didn't find it.
• I checked the LangChain documentation and API reference to see if this feature already exists.
• This is not related to the langchain-community package.

Package (Required)

• langchain
• langchain-openai
• langchain-anthropic
• langchain-classic
• langchain-core
• langchain-model-profiles
• langchain-tests
• langchain-text-splitters
• langchain-chroma
• langchain-deepseek
• langchain-exa
• langchain-fireworks
• langchain-groq
• langchain-huggingface
• langchain-mistralai
• langchain-nomic
• langchain-ollama
• langchain-openrouter
• langchain-perplexity
• langchain-qdrant
• langchain-xai
• Other / not sure / general

Feature Description

I would like LangChain to provide a langchain-a2a-adapters package that wraps remote A2A agents as LangChain tools — mirroring the langchain-mcp-adapters pattern.

langchain-mcp-adapters solved the tool integration problem with a simple dict-based config. Enterprises now need the same for agents — the A2A protocol is the emerging standard for agent-to-agent communication, backed by Google and 150+ organizations.

While LangGraph Platform offers A2A support, there's no open-source adapter for self-hosted deployments.

Use Case

Enterprises are adopting A2A to enable multi-vendor agent interoperability. Teams building with LangChain need to call agents built with other frameworks (AutoGen, CrewAI, Google ADK) as tools within their LangChain agents.

Proposed Solution

A langchain-a2a-adapters package following the same pattern as langchain-mcp-adapters:

from langchain_a2a_adapters.client import MultiAgentA2AClient
from langchain.agents import create_agent

client = MultiAgentA2AClient(
    {
        "research_agent": {
            "url": "http://localhost:9999/",
        },
        "coding_agent": {
            "url": "http://localhost:8888/",
            "headers": {"Authorization": "Bearer TOKEN"},
        }
    }
)

tools = await client.get_tools()
agent = create_agent("anthropic:claude-sonnet-4-20250514", tools)

response = await agent.ainvoke(
    {"messages": [{"role": "user", "content": "Research quantum computing trends"}]}
)

Generated Tool Signature

Each A2A agent tool would have a clear signature:

def research_agent(message: str) -> str:
    """Research agent that can analyze topics and synthesize information.

    Skills:

topic_research: Deep research on any topic
summarization: Synthesize findings into reports
    
    Args:
        message: The task or question to send to the agent.

    Returns:
        The agent's response.
    """

Session/Task Management

The adapter handles A2A task lifecycle internally:

Stateless by default (like langchain-mcp-adapters): Each tool.invoke(message="...") creates a new A2A task, waits for completion, returns the result Stateful sessions (optional): For multi-turn conversations, the tool can accept an optional task_id parameter. If provided, it continues an existing A2A task instead of creating a new one. The task_id would be stored in LangGraph state:

def research_agent(message: str, task_id: str | None = None) -> str:
    """Research agent that can analyze topics and synthesize information.

    Args:
        message: The task or question to send to the agent.
        task_id: Optional. Continue an existing A2A task instead of creating a new one.

    Returns:
        The agent's response.
    """

The returned ToolMessage artifact could include the task_id for the agent to persist in state if multi-turn is needed.

Alternatives Considered

LangGraph Platform A2A endpoint — requires platform deployment, not available for self-hosted open-source users Custom implementations — works but requires manual A2A client management

Additional Context

Related issues:

#32645 (closed — pointed to LangGraph Platform, no open-source adapter)

References:

A2A Protocol: https://google.github.io/A2A/specification/ [lang](https://google.github.io/A2A/specification/)chain-mcp-adapters: https://github.com/langchain-ai/langchain-mcp-adapters A2A Agent Card schema: https://google.github.io/A2A/specification/#agent-card