Hosted MCP support

Author: rm-openai

examples/hosted_mcp/__init__.py

@@ -0,0 +1,61 @@
+import argparse
+import asyncio
+
+from agents import (
+    Agent,
+    HostedMCPTool,
+    MCPToolApprovalFunctionResult,
+    MCPToolApprovalRequest,
+    Runner,
+)
+
+"""This example demonstrates how to use the hosted MCP support in the OpenAI Responses API, with
+approval callbacks."""
+
+
+def approval_callback(request: MCPToolApprovalRequest) -> MCPToolApprovalFunctionResult:
+    answer = input(f"Approve running the tool `{request.data.name}`? (y/n) ")
+    result: MCPToolApprovalFunctionResult = {"approve": answer == "y"}
+    if not result["approve"]:
+        result["reason"] = "User denied"
+    return result
+
+
+async def main(verbose: bool, stream: bool):
+    agent = Agent(
+        name="Assistant",
+        tools=[
+            HostedMCPTool(
+                tool_config={
+                    "type": "mcp",
+                    "server_label": "gitmcp",
+                    "server_url": "https://gitmcp.io/openai/codex",
+                    "require_approval": "always",
+                },
+                on_approval_request=approval_callback,
+            )
+        ],
+    )
+
+    if stream:
+        result = Runner.run_streamed(agent, "Which language is this repo written in?")
+        async for event in result.stream_events():
+            if event.type == "run_item_stream_event":
+                print(f"Got event of type {event.item.__class__.__name__}")
+        print(f"Done streaming; final result: {result.final_output}")
+    else:
+        res = await Runner.run(agent, "Which language is this repo written in?")
+        print(res.final_output)
+
+    if verbose:
+        for item in result.new_items:
+            print(item)
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--verbose", action="store_true", default=False)
+    parser.add_argument("--stream", action="store_true", default=False)
+    args = parser.parse_args()
+
+    asyncio.run(main(args.verbose, args.stream))

examples/hosted_mcp/approvals.py

@@ -0,0 +1,47 @@
+import argparse
+import asyncio
+
+from agents import Agent, HostedMCPTool, Runner
+
+"""This example demonstrates how to use the hosted MCP support in the OpenAI Responses API, with
+approvals not required for any tools. You should only use this for trusted MCP servers."""
+
+
+async def main(verbose: bool, stream: bool):
+    agent = Agent(
+        name="Assistant",
+        tools=[
+            HostedMCPTool(
+                tool_config={
+                    "type": "mcp",
+                    "server_label": "gitmcp",
+                    "server_url": "https://gitmcp.io/openai/codex",
+                    "require_approval": "never",
+                }
+            )
+        ],
+    )
+
+    if stream:
+        result = Runner.run_streamed(agent, "Which language is this repo written in?")
+        async for event in result.stream_events():
+            if event.type == "run_item_stream_event":
+                print(f"Got event of type {event.item.__class__.__name__}")
+        print(f"Done streaming; final result: {result.final_output}")
+    else:
+        res = await Runner.run(agent, "Which language is this repo written in?")
+        print(res.final_output)
+        # The repository is primarily written in multiple languages, including Rust and TypeScript...
+
+    if verbose:
+        for item in result.new_items:
+            print(item)
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--verbose", action="store_true", default=False)
+    parser.add_argument("--stream", action="store_true", default=False)
+    args = parser.parse_args()
+
+    asyncio.run(main(args.verbose, args.stream))

examples/hosted_mcp/simple.py

@@ -58,6 +58,10 @@
     FileSearchTool,
     FunctionTool,
     FunctionToolResult,
+    HostedMCPTool,
+    MCPToolApprovalFunction,
+    MCPToolApprovalFunctionResult,
+    MCPToolApprovalRequest,
     Tool,
     WebSearchTool,
     default_tool_error_function,
@@ -208,6 +212,10 @@ def enable_verbose_stdout_logging():
     "FileSearchTool",
     "Tool",
     "WebSearchTool",
+    "HostedMCPTool",
+    "MCPToolApprovalFunction",
+    "MCPToolApprovalRequest",
+    "MCPToolApprovalFunctionResult",
     "function_tool",
     "Usage",
     "add_trace_processor",

src/agents/__init__.py

@@ -25,7 +25,8 @@
     ActionType,
     ActionWait,
 )
-from openai.types.responses.response_input_param import ComputerCallOutput
+from openai.types.responses.response_input_param import ComputerCallOutput, McpApprovalResponse
+from openai.types.responses.response_output_item import McpApprovalRequest, McpCall, McpListTools
 from openai.types.responses.response_reasoning_item import ResponseReasoningItem
 
 from .agent import Agent, ToolsToFinalOutputResult
@@ -38,6 +39,9 @@
     HandoffCallItem,
     HandoffOutputItem,
     ItemHelpers,
+    MCPApprovalRequestItem,
+    MCPApprovalResponseItem,
+    MCPListToolsItem,
     MessageOutputItem,
     ModelResponse,
     ReasoningItem,
@@ -52,7 +56,14 @@
 from .models.interface import ModelTracing
 from .run_context import RunContextWrapper, TContext
 from .stream_events import RunItemStreamEvent, StreamEvent
-from .tool import ComputerTool, FunctionTool, FunctionToolResult, Tool
+from .tool import (
+    ComputerTool,
+    FunctionTool,
+    FunctionToolResult,
+    HostedMCPTool,
+    MCPToolApprovalRequest,
+    Tool,
+)
 from .tracing import (
     SpanError,
     Trace,
@@ -112,22 +123,30 @@ class ToolRunComputerAction:
     computer_tool: ComputerTool
 
 
+@dataclass
+class ToolRunMCPApprovalRequest:
+    request_item: McpApprovalRequest
+    mcp_tool: HostedMCPTool
+
+
 @dataclass
 class ProcessedResponse:
     new_items: list[RunItem]
     handoffs: list[ToolRunHandoff]
     functions: list[ToolRunFunction]
     computer_actions: list[ToolRunComputerAction]
     tools_used: list[str]  # Names of all tools used, including hosted tools
+    mcp_approval_requests: list[ToolRunMCPApprovalRequest]  # Only requests with callbacks
 
-    def has_tools_to_run(self) -> bool:
+    def has_tools_or_approvals_to_run(self) -> bool:
         # Handoffs, functions and computer actions need local processing
         # Hosted tools have already run, so there's nothing to do.
         return any(
             [
                 self.handoffs,
                 self.functions,
                 self.computer_actions,
+                self.mcp_approval_requests,
             ]
         )
 
@@ -226,7 +245,16 @@ async def execute_tools_and_side_effects(
         new_step_items.extend([result.run_item for result in function_results])
         new_step_items.extend(computer_results)
 
-        # Second, check if there are any handoffs
+        # Next, run the MCP approval requests
+        if processed_response.mcp_approval_requests:
+            approval_results = await cls.execute_mcp_approval_requests(
+                agent=agent,
+                approval_requests=processed_response.mcp_approval_requests,
+                context_wrapper=context_wrapper,
+            )
+            new_step_items.extend(approval_results)
+
+        # Next, check if there are any handoffs
         if run_handoffs := processed_response.handoffs:
             return await cls.execute_handoffs(
                 agent=agent,
@@ -240,7 +268,7 @@ async def execute_tools_and_side_effects(
                 run_config=run_config,
             )
 
-        # Third, we'll check if the tool use should result in a final output
+        # Next, we'll check if the tool use should result in a final output
         check_tool_use = await cls._check_for_final_output_from_tools(
             agent=agent,
             tool_results=function_results,
@@ -295,7 +323,7 @@ async def execute_tools_and_side_effects(
             )
         elif (
             not output_schema or output_schema.is_plain_text()
-        ) and not processed_response.has_tools_to_run():
+        ) and not processed_response.has_tools_or_approvals_to_run():
             return await cls.execute_final_output(
                 agent=agent,
                 original_input=original_input,
@@ -343,10 +371,16 @@ def process_model_response(
         run_handoffs = []
         functions = []
         computer_actions = []
+        mcp_approval_requests = []
         tools_used: list[str] = []
         handoff_map = {handoff.tool_name: handoff for handoff in handoffs}
         function_map = {tool.name: tool for tool in all_tools if isinstance(tool, FunctionTool)}
         computer_tool = next((tool for tool in all_tools if isinstance(tool, ComputerTool)), None)
+        hosted_mcp_server_map = {
+            tool.tool_config["server_label"]: tool
+            for tool in all_tools
+            if isinstance(tool, HostedMCPTool)
+        }
 
         for output in response.output:
             if isinstance(output, ResponseOutputMessage):
@@ -375,6 +409,34 @@ def process_model_response(
                 computer_actions.append(
                     ToolRunComputerAction(tool_call=output, computer_tool=computer_tool)
                 )
+            elif isinstance(output, McpApprovalRequest):
+                items.append(MCPApprovalRequestItem(raw_item=output, agent=agent))
+                if output.server_label not in hosted_mcp_server_map:
+                    _error_tracing.attach_error_to_current_span(
+                        SpanError(
+                            message="MCP server label not found",
+                            data={"server_label": output.server_label},
+                        )
+                    )
+                    raise ModelBehaviorError(f"MCP server label {output.server_label} not found")
+                else:
+                    server = hosted_mcp_server_map[output.server_label]
+                    if server.on_approval_request:
+                        mcp_approval_requests.append(
+                            ToolRunMCPApprovalRequest(
+                                request_item=output,
+                                mcp_tool=server,
+                            )
+                        )
+                    else:
+                        logger.warning(
+                            f"MCP server {output.server_label} has no on_approval_request hook"
+                        )
+            elif isinstance(output, McpListTools):
+                items.append(MCPListToolsItem(raw_item=output, agent=agent))
+            elif isinstance(output, McpCall):
+                items.append(ToolCallItem(raw_item=output, agent=agent))
+                tools_used.append(output.name)
             elif not isinstance(output, ResponseFunctionToolCall):
                 logger.warning(f"Unexpected output type, ignoring: {type(output)}")
                 continue
@@ -417,6 +479,7 @@ def process_model_response(
             functions=functions,
             computer_actions=computer_actions,
             tools_used=tools_used,
+            mcp_approval_requests=mcp_approval_requests,
         )
 
     @classmethod
@@ -643,6 +706,40 @@ async def execute_handoffs(
             next_step=NextStepHandoff(new_agent),
         )
 
+    @classmethod
+    async def execute_mcp_approval_requests(
+        cls,
+        *,
+        agent: Agent[TContext],
+        approval_requests: list[ToolRunMCPApprovalRequest],
+        context_wrapper: RunContextWrapper[TContext],
+    ) -> list[RunItem]:
+        async def run_single_approval(approval_request: ToolRunMCPApprovalRequest) -> RunItem:
+            callback = approval_request.mcp_tool.on_approval_request
+            assert callback is not None, "Callback is required for MCP approval requests"
+            maybe_awaitable_result = callback(
+                MCPToolApprovalRequest(context_wrapper, approval_request.request_item)
+            )
+            if inspect.isawaitable(maybe_awaitable_result):
+                result = await maybe_awaitable_result
+            else:
+                result = maybe_awaitable_result
+            reason = result.get("reason", None)
+            raw_item: McpApprovalResponse = {
+                "approval_request_id": approval_request.request_item.id,
+                "approve": result["approve"],
+                "type": "mcp_approval_response",
+            }
+            if not result["approve"] and reason:
+                raw_item["reason"] = reason
+            return MCPApprovalResponseItem(
+                raw_item=raw_item,
+                agent=agent,
+            )
+
+        tasks = [run_single_approval(approval_request) for approval_request in approval_requests]
+        return await asyncio.gather(*tasks)
+
     @classmethod
     async def execute_final_output(
         cls,
@@ -727,6 +824,11 @@ def stream_step_result_to_queue(
                 event = RunItemStreamEvent(item=item, name="tool_output")
             elif isinstance(item, ReasoningItem):
                 event = RunItemStreamEvent(item=item, name="reasoning_item_created")
+            elif isinstance(item, MCPApprovalRequestItem):
+                event = RunItemStreamEvent(item=item, name="mcp_approval_requested")
+            elif isinstance(item, MCPListToolsItem):
+                event = RunItemStreamEvent(item=item, name="mcp_list_tools")
+
             else:
                 logger.warning(f"Unexpected item type: {type(item)}")
                 event = None