Address automated review comments: span error for ToolExecutionException, README prereqs, and AsyncClient lifecycle

Co-authored-by: TaoChenOSU <12570346+TaoChenOSU@users.noreply.github.com>

python/packages/core/agent_framework/_mcp.py

@@ -950,6 +950,7 @@ class MCPTool:
 
         with get_mcp_call_span(span_attributes) as span:
             # Try the operation, reconnecting once if the connection is closed
+            _span_error_set = False
             for attempt in range(2):
                 try:
                     # Capture the JSON-RPC request ID before the call is made.
@@ -975,9 +976,13 @@ class MCPTool:
                         error_msg = text or str(parsed)
                         span.set_attribute(OtelAttr.ERROR_TYPE, "ToolError")
                         span.set_status(trace.StatusCode.ERROR, error_msg)
+                        _span_error_set = True
                         raise ToolExecutionException(error_msg)
                     return parser(result)
-                except ToolExecutionException:
+                except ToolExecutionException as ex:
+                    if not _span_error_set:
+                        span.set_attribute(OtelAttr.ERROR_TYPE, type(ex).__name__)
+                        span.set_status(trace.StatusCode.ERROR, str(ex))
                     raise
                 except ClosedResourceError as cl_ex:
                     if attempt == 0:

python/samples/02-agents/mcp/README.md

@@ -17,8 +17,7 @@ The Model Context Protocol (MCP) is an open standard for connecting AI agents to
 
 ## Prerequisites
 
-- `OPENAI_API_KEY` environment variable
-- `OPENAI_RESPONSES_MODEL_ID` environment variable
+Each sample requires its own set of environment variables. See below for details.
 
 For `mcp_github_pat_openai_responses.py`:
 - `GITHUB_PAT` - Your GitHub Personal Access Token (create at https://github.com/settings/tokens)

python/samples/02-agents/mcp/mcp_github_pat_azure_chat.py

@@ -46,24 +46,23 @@ async def github_mcp_example() -> None:
         "Authorization": f"Bearer {github_pat}",
     }
 
-    # 4. Create HTTP client with authentication headers
-    http_client = AsyncClient(headers=auth_headers)
-
-    # 5. Create Azure OpenAI Chat Client
+    # 4. Create Azure OpenAI Chat Client
     # For authentication, run `az login` command in terminal or replace AzureCliCredential
     # with your preferred authentication option (e.g., api_key parameter)
     client = AzureOpenAIChatClient(credential=AzureCliCredential())
 
-    # 6. Create MCP tool for GitHub with PAT authentication
-    # The MCPStreamableHTTPTool manages the connection to the MCP server and makes its tools available
-    async with MCPStreamableHTTPTool(
+    # 5. Create HTTP client with authentication headers and MCP tool for GitHub with PAT authentication.
+    # The MCPStreamableHTTPTool manages the connection to the MCP server and makes its tools available.
+    # The HTTP client is used to pass the authentication headers to the MCP server and is closed when
+    # the context manager exits.
+    async with AsyncClient(headers=auth_headers) as http_client, MCPStreamableHTTPTool(
         name="GitHub",
         description="GitHub MCP server for interacting with GitHub repositories, issues, and more",
         url="https://api.githubcopilot.com/mcp/",
         http_client=http_client,  # Pass HTTP client with authentication headers
         approval_mode="never_require",  # For sample brevity; use "always_require" in production
     ) as github_mcp_tool:
-        # 7. Create agent with the GitHub MCP tool
+        # 6. Create agent with the GitHub MCP tool
         agent = client.as_agent(
             instructions=(
                 "You are a helpful assistant that can help users interact with GitHub. "