LangChain Model Context Protocol (MCP) 适配器¶

Client for connecting to multiple MCP servers and loading LangChain-compatible resources.

This module provides the MultiServerMCPClient class for managing connections to multiple MCP servers and loading tools, prompts, and resources from them.

Classes:

Name	Description
`MultiServerMCPClient`	Client for connecting to multiple MCP servers and loading LangChain-compatible tools, prompts and resources from them.

MultiServerMCPClient ¶

Client for connecting to multiple MCP servers and loading LangChain-compatible tools, prompts and resources from them.

Methods:

Name	Description
`__init__`	Initialize a MultiServerMCPClient with MCP servers connections.
`session`	Connect to an MCP server and initialize a session.
`get_tools`	Get a list of all tools from all connected servers.
`get_prompt`	Get a prompt from a given MCP server.
`get_resources`	Get resources from a given MCP server.
`__aenter__`	Async context manager entry point.
`__aexit__`	Async context manager exit point.

init ¶

__init__(
    connections: dict[str, Connection] | None = None,
) -> None

Initialize a MultiServerMCPClient with MCP servers connections.

Parameters:

Name	Type	Description	Default
`connections`	`dict[str, Connection] \| None`	A dictionary mapping server names to connection configurations. If None, no initial connections are established.	`None`

Example: basic usage (starting a new session on each tool call)

from langchain_mcp_adapters.client import MultiServerMCPClient

client = MultiServerMCPClient(
    {
        "math": {
            "command": "python",
            # Make sure to update to the full absolute path to your math_server.py file
            "args": ["/path/to/math_server.py"],
            "transport": "stdio",
        },
        "weather": {
            # Make sure you start your weather server on port 8000
            "url": "http://localhost:8000/mcp",
            "transport": "streamable_http",
        }
    }
)
all_tools = await client.get_tools()

Example: explicitly starting a session

from langchain_mcp_adapters.client import MultiServerMCPClient
from langchain_mcp_adapters.tools import load_mcp_tools

client = MultiServerMCPClient({...})
async with client.session("math") as session:
    tools = await load_mcp_tools(session)

session `async` ¶

session(
    server_name: str, *, auto_initialize: bool = True
) -> AsyncIterator[ClientSession]

Connect to an MCP server and initialize a session.

Parameters:

Name	Type	Description	Default
`server_name`	`str`	Name to identify this server connection	required
`auto_initialize`	`bool`	Whether to automatically initialize the session	`True`

Raises:

Type	Description
`ValueError`	If the server name is not found in the connections

Yields:

Type	Description
`AsyncIterator[ClientSession]`	An initialized ClientSession

get_tools `async` ¶

get_tools(
    *, server_name: str | None = None
) -> list[BaseTool]

Get a list of all tools from all connected servers.

Parameters:

Name	Type	Description	Default
`server_name`	`str \| None`	Optional name of the server to get tools from. If None, all tools from all servers will be returned (default).	`None`

NOTE: a new session will be created for each tool call

Returns:

Type	Description
`list[BaseTool]`	A list of LangChain tools

get_prompt `async` ¶

get_prompt(
    server_name: str,
    prompt_name: str,
    *,
    arguments: dict[str, Any] | None = None
) -> list[HumanMessage | AIMessage]

Get a prompt from a given MCP server.

get_resources `async` ¶

get_resources(
    server_name: str, *, uris: str | list[str] | None = None
) -> list[Blob]

Get resources from a given MCP server.

Parameters:

Name	Type	Description	Default
`server_name`	`str`	Name of the server to get resources from	required
`uris`	`str \| list[str] \| None`	Optional resource URI or list of URIs to load. If not provided, all resources will be loaded.	`None`

Returns:

Type	Description
`list[Blob]`	A list of LangChain Blobs

aenter `async` ¶

__aenter__() -> MultiServerMCPClient

Async context manager entry point.

Raises:

Type	Description
`NotImplementedError`	Context manager support has been removed.

aexit ¶

__aexit__(
    exc_type: type[BaseException] | None,
    exc_val: BaseException | None,
    exc_tb: TracebackType | None,
) -> None

Async context manager exit point.

Parameters:

Name	Type	Description	Default
`exc_type`	`type[BaseException] \| None`	Exception type if an exception occurred.	required
`exc_val`	`BaseException \| None`	Exception value if an exception occurred.	required
`exc_tb`	`TracebackType \| None`	Exception traceback if an exception occurred.	required

Raises:

Type	Description
`NotImplementedError`	Context manager support has been removed.

Tools adapter for converting MCP tools to LangChain tools.

This module provides functionality to convert MCP tools into LangChain-compatible tools, handle tool execution, and manage tool conversion between the two formats.

Functions:

Name	Description
`load_mcp_tools`	Load all available MCP tools and convert them to LangChain tools.

load_mcp_tools `async` ¶

load_mcp_tools(
    session: ClientSession | None,
    *,
    connection: Connection | None = None
) -> list[BaseTool]

Load all available MCP tools and convert them to LangChain tools.

Parameters:

Name	Type	Description	Default
`session`	`ClientSession \| None`	The MCP client session. If None, connection must be provided.	required
`connection`	`Connection \| None`	Connection config to create a new session if session is None.	`None`

Returns:

Type	Description
`list[BaseTool]`	List of LangChain tools. Tool annotations are returned as part
`list[BaseTool]`	of the tool metadata object.

Raises:

Type	Description
`ValueError`	If neither session nor connection is provided.

Prompts adapter for converting MCP prompts to LangChain messages.

This module provides functionality to convert MCP prompt messages into LangChain message objects, handling both user and assistant message types.

Functions:

Name	Description
`load_mcp_prompt`	Load MCP prompt and convert to LangChain messages.

load_mcp_prompt `async` ¶

load_mcp_prompt(
    session: ClientSession,
    name: str,
    *,
    arguments: dict[str, Any] | None = None
) -> list[HumanMessage | AIMessage]

Load MCP prompt and convert to LangChain messages.

Parameters:

Name	Type	Description	Default
`session`	`ClientSession`	The MCP client session.	required
`name`	`str`	Name of the prompt to load.	required
`arguments`	`dict[str, Any] \| None`	Optional arguments to pass to the prompt.	`None`

Returns:

Type	Description
`list[HumanMessage \| AIMessage]`	A list of LangChain messages converted from the MCP prompt.

Resources adapter for converting MCP resources to LangChain Blobs.

This module provides functionality to convert MCP resources into LangChain Blob objects, handling both text and binary resource content types.

Functions:

Name	Description
`load_mcp_resources`	Load MCP resources and convert them to LangChain Blobs.

load_mcp_resources `async` ¶

load_mcp_resources(
    session: ClientSession,
    *,
    uris: str | list[str] | None = None
) -> list[Blob]

Load MCP resources and convert them to LangChain Blobs.

Parameters:

Name	Type	Description	Default
`session`	`ClientSession`	MCP client session.	required
`uris`	`str \| list[str] \| None`	List of URIs to load. If None, all resources will be loaded. Note: Dynamic resources will NOT be loaded when None is specified, as they require parameters and are ignored by the MCP SDK's session.list_resources() method.	`None`

Returns:

Type	Description
`list[Blob]`	A list of LangChain Blobs.

Raises:

Type	Description
`RuntimeError`	If an error occurs while fetching a resource.

LangChain Model Context Protocol (MCP) 适配器¶

MultiServerMCPClient ¶

__init__ ¶

session async ¶

get_tools async ¶

get_prompt async ¶

get_resources async ¶

__aenter__ async ¶

__aexit__ ¶

load_mcp_tools async ¶

load_mcp_prompt async ¶

load_mcp_resources async ¶

init ¶

session `async` ¶

get_tools `async` ¶

get_prompt `async` ¶

get_resources `async` ¶

aenter `async` ¶

aexit ¶

load_mcp_tools `async` ¶

load_mcp_prompt `async` ¶

load_mcp_resources `async` ¶