McpToolRegistrationService Class

Provides MCP tool registration services for Agent Framework agents.

This service handles registration and management of MCP (Model Context Protocol) tool servers with Agent Framework agents.

Constructor

McpToolRegistrationService()

Parameters

Name	Description
logger	Default value: None

Methods

__init__	Initialize the MCP Tool Registration Service for Agent Framework.
__new__
add_tool_servers_to_agent	Add MCP tool servers to a RawAgent (mirrors .NET implementation).
cleanup	Clean up any resources used by the service.
send_chat_history_from_store	Send chat history from a HistoryProvider to the MCP platform. This is a convenience method that extracts messages from the store and delegates to send_chat_history_messages().
send_chat_history_messages	Send chat history messages to the MCP platform for real-time threat protection. This is the primary implementation method that handles message conversion and delegation to the core tooling service. Note Even if chat_messages is empty or all messages are filtered during conversion, the request will still be sent to the MCP platform. This ensures the user message from turn_context.activity.text is registered correctly for real-time threat protection.

init

Initialize the MCP Tool Registration Service for Agent Framework.

__init__(logger: Logger | None = None)

Parameters

Name	Description
logger	Logger \| None Logger instance for logging operations. Default value: None

new

__new__(**kwargs)

add_tool_servers_to_agent

Add MCP tool servers to a RawAgent (mirrors .NET implementation).

async add_tool_servers_to_agent(chat_client: OpenAIChatClient, agent_instructions: str, initial_tools: List[object], auth: Authorization, auth_handler_name: str, turn_context: TurnContext, auth_token: str | None = None) -> RawAgent

Parameters

Name	Description
chat_client Required	<xref:OpenAIChatClient> The chat client instance (OpenAIChatClient supports both OpenAI and Azure OpenAI)
agent_instructions Required	str Instructions for the agent behavior
initial_tools Required	List[object] List of initial tools to add to the agent
auth Required	<xref:Authorization> Authorization context for token exchange
auth_handler_name Required	str Name of the authorization handler.
turn_context Required	<xref:TurnContext> Turn context for the operation
auth_token	Optional[str] Optional bearer token for authentication Default value: None

Returns

Type	Description
<xref:RawAgent>	RawAgent instance with MCP tools registered.

Exceptions

Type	Description
Exception	If agent creation fails.

cleanup

Clean up any resources used by the service.

async cleanup()

send_chat_history_from_store

Send chat history from a HistoryProvider to the MCP platform.

This is a convenience method that extracts messages from the store and delegates to send_chat_history_messages().

async send_chat_history_from_store(chat_message_store: HistoryProvider, turn_context: TurnContext, tool_options: ToolOptions | None = None) -> OperationResult

Parameters

Name	Description
chat_message_store Required	<xref:agent_framework._sessions.HistoryProvider> HistoryProvider containing the conversation history.
turn_context Required	TurnContext TurnContext from the Agents SDK containing conversation info.
tool_options	ToolOptions \| None Optional configuration for the request. Default value: None

Returns

Type	Description
OperationResult	OperationResult indicating success or failure of the operation.

Exceptions

Type	Description
ValueError	If chat_message_store or turn_context is None.

Examples


>>> service = McpToolRegistrationService()
>>> result = await service.send_chat_history_from_store(
...     thread.chat_message_store, turn_context
... )

send_chat_history_messages

Send chat history messages to the MCP platform for real-time threat protection.

This is the primary implementation method that handles message conversion and delegation to the core tooling service.

Note

Even if chat_messages is empty or all messages are filtered during

conversion, the request will still be sent to the MCP platform. This

ensures the user message from turn_context.activity.text is registered

correctly for real-time threat protection.

async send_chat_history_messages(chat_messages: Sequence[Message], turn_context: TurnContext, tool_options: ToolOptions | None = None) -> OperationResult

Parameters

Name	Description
chat_messages Required	Sequence[<xref:agent_framework._types.Message>] Sequence of Agent Framework Message objects to send. Can be empty - the request will still be sent to register the user message from turn_context.activity.text.
turn_context Required	TurnContext TurnContext from the Agents SDK containing conversation info.
tool_options	ToolOptions \| None Optional configuration for the request. Defaults to AgentFramework-specific options if not provided. Default value: None

Returns

Type	Description
OperationResult	OperationResult indicating success or failure of the operation.

Exceptions

Type	Description
ValueError	If chat_messages or turn_context is None.

Examples


>>> service = McpToolRegistrationService()
>>> messages = [Message(role="user", text="Hello")]
>>> result = await service.send_chat_history_messages(messages, turn_context)
>>> if result.succeeded:
...     print("Chat history sent successfully")