fix: normalize restored tool results

src/strands/session/repository_session_manager.py

@@ -282,25 +282,39 @@ def _fix_broken_tool_use(self, messages: list[Message]) -> list[Message]:
                     ]
 
                     # Check if there are more messages after the current toolUse message
-                    tool_result_ids = [
-                        content["toolResult"]["toolUseId"]
-                        for content in messages[index + 1]["content"]
-                        if "toolResult" in content
-                    ]
+                    next_message = messages[index + 1]
+                    tool_use_id_set = set(tool_use_ids)
+                    tool_result_ids = []
+                    filtered_content = []
+                    removed_tool_results = False
+
+                    for content in next_message["content"]:
+                        if "toolResult" not in content:
+                            filtered_content.append(content)
+                            continue
+
+                        tool_use_id = content["toolResult"]["toolUseId"]
+                        if tool_use_id in tool_use_id_set and tool_use_id not in tool_result_ids:
+                            tool_result_ids.append(tool_use_id)
+                            filtered_content.append(content)
+                        else:
+                            removed_tool_results = True
 
-                    missing_tool_use_ids = list(set(tool_use_ids) - set(tool_result_ids))
+                    missing_tool_use_ids = [
+                        tool_use_id for tool_use_id in tool_use_ids if tool_use_id not in tool_result_ids
+                    ]
                     # If there are missing tool use ids, that means the messages history is broken
-                    if missing_tool_use_ids:
+                    if missing_tool_use_ids or removed_tool_results:
                         logger.warning(
                             "Session message history has an orphaned toolUse with no toolResult. "
                             "Adding toolResult content blocks to create valid conversation."
                         )
                         # Create the missing toolResult content blocks
                         missing_content_blocks = generate_missing_tool_result_content(missing_tool_use_ids)
 
-                        if tool_result_ids:
-                            # If there were any toolResult ids, that means only some of the content blocks are missing
-                            messages[index + 1]["content"].extend(missing_content_blocks)
+                        if tool_result_ids or removed_tool_results:
+                            # Keep only toolResults that match the previous assistant toolUse blocks.
+                            next_message["content"] = filtered_content + missing_content_blocks
                         else:
                             # The message following the toolUse was not a toolResult, so lets insert it
                             messages.insert(index + 1, {"role": "user", "content": missing_content_blocks})

tests/strands/session/test_repository_session_manager.py

@@ -370,6 +370,33 @@ def test_fix_broken_tool_use_extends_partial_tool_results(existing_session_manag
     assert missing_result["toolResult"]["content"][0]["text"] == "Tool was interrupted."
 
 
+def test_fix_broken_tool_use_drops_extra_tool_results(session_manager):
+    messages = [
+        {
+            "role": "assistant",
+            "content": [
+                {"toolUse": {"toolUseId": "complete-123", "name": "test_tool", "input": {"input": "test1"}}},
+                {"toolUse": {"toolUseId": "missing-456", "name": "test_tool", "input": {"input": "test2"}}},
+            ],
+        },
+        {
+            "role": "user",
+            "content": [
+                {"toolResult": {"toolUseId": "complete-123", "status": "success", "content": [{"text": "ok"}]}},
+                {"toolResult": {"toolUseId": "stale-789", "status": "error", "content": [{"text": "old"}]}},
+                {"toolResult": {"toolUseId": "complete-123", "status": "success", "content": [{"text": "dup"}]}},
+            ],
+        },
+    ]
+
+    fixed_messages = session_manager._fix_broken_tool_use(messages)
+
+    tool_results = [content["toolResult"] for content in fixed_messages[1]["content"]]
+    assert [tool_result["toolUseId"] for tool_result in tool_results] == ["complete-123", "missing-456"]
+    assert tool_results[0]["content"] == [{"text": "ok"}]
+    assert tool_results[1]["status"] == "error"
+
+
 def test_fix_broken_tool_use_handles_multiple_orphaned_tools(existing_session_manager):
     """Test fixing multiple orphaned toolUse messages."""