Python: Add regression tests for MCP tool result event projection in GitHubCopilotAgent

python/packages/core/agent_framework/_skills.py

@@ -2134,9 +2134,7 @@ async def _run_script(
             ),
             FunctionTool(
                 name="read_skill_resource",
-                description=(
-                    "Reads a resource associated with a skill, such as references, assets, or dynamic data."
-                ),
+                description=("Reads a resource associated with a skill, such as references, assets, or dynamic data."),
                 func=_read_resource,
                 input_model={
                     "type": "object",
@@ -2173,8 +2171,7 @@ async def _run_script(
                                     "type": "object",
                                     "additionalProperties": True,
                                     "description": (
-                                        "Named arguments as key-value pairs "
-                                        '(e.g. {"length": 24, "uppercase": true}).'
+                                        'Named arguments as key-value pairs (e.g. {"length": 24, "uppercase": true}).'
                                     ),
                                 },
                                 {

python/packages/core/tests/core/test_skills.py

@@ -4086,8 +4086,8 @@ async def test_full_skill_content_contains_scripts(self) -> None:
 
     async def test_content_is_cached(self) -> None:
         skill = _MinimalClassSkill()
-        content1 = (await skill.get_content())
-        content2 = (await skill.get_content())
+        content1 = await skill.get_content()
+        content2 = await skill.get_content()
         assert content1 is content2
 
     def test_resources_are_lazy_cached(self) -> None:
@@ -5587,8 +5587,8 @@ class TestInlineSkillContentCaching:
     async def test_content_cached_after_first_access(self) -> None:
         """InlineSkill.content returns the same object on subsequent accesses."""
         skill = InlineSkill(frontmatter=SkillFrontmatter(name="test-skill", description="Test"), instructions="Body")
-        first = (await skill.get_content())
-        second = (await skill.get_content())
+        first = await skill.get_content()
+        second = await skill.get_content()
         assert first is second  # Same object (cached)
         assert "<name>test-skill</name>" in first
 

python/packages/github_copilot/tests/test_github_copilot_agent.py

@@ -920,6 +920,87 @@ def mock_on(handler: Any) -> Any:
         assert responses[3].role == "assistant"
         assert responses[3].contents[0].type == "text"
 
+    async def test_run_streaming_multiple_tool_calls_with_json_results(
+        self,
+        mock_client: MagicMock,
+        mock_session: MagicMock,
+        session_idle_event: SessionEvent,
+    ) -> None:
+        """Test that multiple tool calls with JSON results are all projected correctly."""
+        json_payloads = [
+            '{"users": [{"id": "u1", "name": "Alice"}, {"id": "u2", "name": "Bob"}]}',
+            '{"tickets": [{"id": "INC001", "state": "open", "short_description": "VPN issue"}]}',
+            '{"groups": [{"id": "g1", "displayName": "IT Admins", "members": 42}]}',
+        ]
+        tool_names = ["msgraph_list_users", "sn_query_table", "msgraph_list_groups"]
+
+        events: list[SessionEvent] = []
+        for i, (name, payload) in enumerate(zip(tool_names, json_payloads)):
+            call_id = f"call_{i:03d}"
+
+            # Tool start
+            start_data = MagicMock()
+            start_data.tool_call_id = call_id
+            start_data.tool_name = name
+            start_data.arguments = {"query": f"test_{i}"}
+            events.append(
+                SessionEvent(
+                    data=start_data,
+                    id=uuid4(),
+                    timestamp=datetime.now(timezone.utc),
+                    type=SessionEventType.TOOL_EXECUTION_START,
+                )
+            )
+
+            # Tool complete
+            complete_data = MagicMock()
+            complete_data.tool_call_id = call_id
+            complete_data.result = ToolExecutionCompleteResult(content=payload)
+            complete_data.success = True
+            complete_data.error = None
+            events.append(
+                SessionEvent(
+                    data=complete_data,
+                    id=uuid4(),
+                    timestamp=datetime.now(timezone.utc),
+                    type=SessionEventType.TOOL_EXECUTION_COMPLETE,
+                )
+            )
+
+        events.append(session_idle_event)
+
+        def mock_on(handler: Any) -> Any:
+            for event in events:
+                handler(event)
+            return lambda: None
+
+        mock_session.on = mock_on
+
+        agent = GitHubCopilotAgent(client=mock_client)
+        responses: list[AgentResponseUpdate] = []
+        async for update in agent.run("List users, tickets, and groups", stream=True):
+            responses.append(update)
+
+        # Should have 3 tool calls + 3 tool results = 6 updates
+        assert len(responses) == 6
+
+        # Verify each pair: function_call followed by function_result
+        for i in range(3):
+            call_update = responses[i * 2]
+            result_update = responses[i * 2 + 1]
+            call_id = f"call_{i:03d}"
+
+            assert call_update.role == "assistant"
+            assert call_update.contents[0].type == "function_call"
+            assert call_update.contents[0].call_id == call_id
+            assert call_update.contents[0].name == tool_names[i]
+
+            assert result_update.role == "tool"
+            assert result_update.contents[0].type == "function_result"
+            assert result_update.contents[0].call_id == call_id
+            assert result_update.contents[0].result == json_payloads[i]
+            assert result_update.contents[0].exception is None
+
 
 class TestGitHubCopilotAgentSessionManagement:
     """Test cases for session management."""

python/samples/02-agents/harness/harness_research.py

@@ -109,7 +109,10 @@ async def main() -> None:
                         print(f"\n  [calling tool: {content.name}]", flush=True)
                         print("  ", end="", flush=True)
                     # Show web search activity when the result arrives with action details.
-                    elif content.type in ("search_tool_call", "search_tool_result") and getattr(content, "tool_name", None) == "web_search":
+                    elif (
+                        content.type in ("search_tool_call", "search_tool_result")
+                        and getattr(content, "tool_name", None) == "web_search"
+                    ):
                         action = None
                         if content.type == "search_tool_result" and isinstance(content.result, dict):
                             action = content.result.get("action", {})