diff --git a/Sensor/README.md b/Sensor/README.md
index bd81ecf..1daa2df 100644
--- a/Sensor/README.md
+++ b/Sensor/README.md
@@ -202,6 +202,10 @@ Each parsed session produces an `AgentEvent` with the following structure:
 }
 ```
 
+Tool arguments and results can contain source code, credentials, or other sensitive
+content copied from the local environment. Treat sensor output as sensitive data and
+review it before sharing.
+
 ### `session_context`
 
 Parsers that can recover session-level configuration attach it under
diff --git a/Sensor/adr_sensor/parsers/cursor_parser.py b/Sensor/adr_sensor/parsers/cursor_parser.py
index 068ad76..7bb9053 100644
--- a/Sensor/adr_sensor/parsers/cursor_parser.py
+++ b/Sensor/adr_sensor/parsers/cursor_parser.py
@@ -286,11 +286,9 @@ def extract_tools_from_bubble(self, bubble: Dict[str, Any]) -> List[ToolUsage]:
                 status = tool_data.get("status", "unknown")
                 error = tool_data.get("error", "")
 
-                result = None
-                if tool_name == "list_dir":
-                    result = tool_data.get("result", "")
-                    if result and isinstance(result, str):
-                        result = truncate_middle(result.strip(), max_length=1000, edge_chars=400)
+                result = tool_data.get("result")
+                if result is not None:
+                    result = truncate_middle(str(result), max_length=1000, edge_chars=400)
 
                 tool = ToolUsage(
                     tool_name=tool_name,
diff --git a/Sensor/tests/test_parsers.py b/Sensor/tests/test_parsers.py
index 4112a14..2626341 100644
--- a/Sensor/tests/test_parsers.py
+++ b/Sensor/tests/test_parsers.py
@@ -435,6 +435,47 @@ def test_parse_no_directory(self):
         assert entries == []
 
 
+class TestCursorParser:
+    @pytest.mark.parametrize("tool_name", ["list_dir", "read_file", "run_command"])
+    def test_extracts_result_for_every_tool(self, tool_name):
+        tools = CursorParser().extract_tools_from_bubble(
+            {"toolFormerData": {"name": tool_name, "params": {}, "result": "completed"}}
+        )
+
+        assert len(tools) == 1
+        assert tools[0].tool_name == tool_name
+        assert tools[0].result == "completed"
+
+    def test_absent_result_remains_none(self):
+        tools = CursorParser().extract_tools_from_bubble(
+            {"toolFormerData": {"name": "read_file", "params": {}}}
+        )
+
+        assert tools[0].result is None
+
+    def test_result_whitespace_is_preserved(self):
+        tools = CursorParser().extract_tools_from_bubble(
+            {"toolFormerData": {"name": "search", "params": {}, "result": " \n  first\nsecond\t "}}
+        )
+
+        assert tools[0].result == " \n  first\nsecond\t "
+
+    def test_non_string_result_is_converted_to_string(self):
+        tools = CursorParser().extract_tools_from_bubble(
+            {"toolFormerData": {"name": "count", "params": {}, "result": 3}}
+        )
+
+        assert tools[0].result == "3"
+
+    def test_result_is_middle_truncated_at_1000_characters(self):
+        result = "a" * 600 + "z" * 600
+        tools = CursorParser().extract_tools_from_bubble(
+            {"toolFormerData": {"name": "read_file", "params": {}, "result": result}}
+        )
+
+        assert tools[0].result == "a" * 400 + "... [truncated 400 chars] ..." + "z" * 400
+
+
 def _build_warp_db(db_path: Path, conversations: list) -> None:
     """Create a synthetic warp.sqlite matching the schema WarpParser queries.