diff --git a/cecli/helpers/conversation/integration.py b/cecli/helpers/conversation/integration.py
index f5c046728e2..f649689323d 100644
--- a/cecli/helpers/conversation/integration.py
+++ b/cecli/helpers/conversation/integration.py
@@ -201,9 +201,19 @@ def add_randomized_cta(self) -> None:
             ]
         )
 
+        user_fidelity = random.choice(
+            [
+                "Be mindful of any instructions given, prioritizing the latest.",
+                "Respect all established constraints.",
+                "Please stay on task and stick closely to my guidance.",
+                "Keep my explicit intent in mind.",
+                "Stay focused on our goals and the scope of our concerns.",
+            ]
+        )
+
         msg = dict(
             role="user",
-            content="System Message:\n\n" + message,
+            content=f"System Message:\n\n{message}\n{user_fidelity}",
         )
 
         ConversationService.get_manager(coder).add_message(
diff --git a/cecli/prompts/agent.yml b/cecli/prompts/agent.yml
index 988538838a5..6ed6e1566b7 100644
--- a/cecli/prompts/agent.yml
+++ b/cecli/prompts/agent.yml
@@ -46,7 +46,7 @@ main_system: |
   2. **Explore**: Use discovery tools (`ExploreCode`, `Grep`, `Ls`) to research and gather understanding for you task. Modify search terms when errors are encountered.
   3. **Execute**: Mark files as editable with `ContextManager` before attempting edits.  Proactively use skills if they are available. Review diff outputs after edit to ensure the proper changes were made.
   4. **Verify & Recover**: If an edit fails or introduces linting errors, use `UndoChange` immediately.
-  5. **Yield**: Use the `Yield` tool only after verifying the solution. Briefly summarize the changes for the user.
+  5. **Yield**: Use the `Yield` tool after accomplishing the goal and verifying any changes made. Provide helpful summaries of any changes.
 
   ## Todo List Management
   - Break complex goals into meaningful sub-tasks so the problem remains tractable
@@ -57,12 +57,12 @@ main_system: |
   </context>
 
   Use the `.cecli/temp` directory for all temporary, test, or scratch files.
-  Always reply to the user in {language}.
+  Always reply in {language}.
 
 system_reminder: |
   <context name="critical_reminders">
   ## Operational Rules
-  - **Scope**: No unrequested refactors. Avoid full-file rewrites.
+  - **Scope**: No unrequested refactors. Avoid full-file rewrites. Only modify what you are asked to.
   - **Hygiene**: Use `ContextManager`/`RemoveSkill` to evict unneeded files/skills immediately after use.
   - **Outputs**: Tool calls trigger turns. Never include tool syntax in final user summaries.
   - **Sandbox**: Perform all verification and temp logic in `.cecli/temp`.
diff --git a/cecli/prompts/subagent.yml b/cecli/prompts/subagent.yml
index 1339499995c..7786b0f6aa9 100644
--- a/cecli/prompts/subagent.yml
+++ b/cecli/prompts/subagent.yml
@@ -31,7 +31,7 @@ main_system: |
   2. **Explore**: Use discovery tools (`ExploreCode`, `Grep`, `Ls`) to research and gather understanding for you task. Modify search terms when errors are encountered.
   3. **Execute**: Mark files as editable with `ContextManager` before attempting edits.  Proactively use skills if they are available. Review diff outputs after edit to ensure the proper changes were made.
   4. **Verify & Recover**: If an edit fails or introduces linting errors, use `UndoChange` immediately.
-  5. **Yield**: Use the `Yield` tool only after verifying the solution. Briefly summarize the changes for the user.
+  5. **Yield**: Use the `Yield` tool after accomplishing the goal and verifying any changes made. Provide helpful summaries of any changes.
 
   ## Todo List Management
   - Break complex goals into meaningful sub-tasks so the problem remains tractable
@@ -42,12 +42,12 @@ main_system: |
   </context>
 
   Use the `.cecli/temp` directory for all temporary, test, or scratch files.
-  Always reply to the user in {language}.
+  Always reply in {language}.
 
 system_reminder: |
   <context name="critical_reminders">
   ## Operational Rules
-  - **Scope**: No unrequested refactors. Avoid full-file rewrites.
+  - **Scope**: No unrequested refactors. Avoid full-file rewrites. Only modify what you are asked to.
   - **Hygiene**: Use `ContextManager`/`RemoveSkill` to evict unneeded files/skills immediately after use.
   - **Outputs**: Tool calls trigger turns. Never include tool syntax in final user summaries.
   - **Sandbox**: Perform all verification and temp logic in `.cecli/temp`.
diff --git a/cecli/tui/app.py b/cecli/tui/app.py
index 9d151bff074..2613c091336 100644
--- a/cecli/tui/app.py
+++ b/cecli/tui/app.py
@@ -722,9 +722,9 @@ def update_spinner(self, msg, agent_name: str | None = None):
 
     def show_error(self, message, agent_name: str | None = None):
         """Show an error message in the status bar."""
-        self.status_bar.show_notification(
-            message, severity="error", timeout=5, agent_name=agent_name
-        )
+        status_bar = self.query_one("#status-bar", StatusBar)
+
+        status_bar.show_notification(message, severity="error", timeout=5, agent_name=agent_name)
 
     def on_resize(self) -> None:
         file_list = self.query_one("#file-list", FileList)
diff --git a/tests/tui/test_app.py b/tests/tui/test_app.py
index 5d008b93ad3..ec68cfd657e 100644
--- a/tests/tui/test_app.py
+++ b/tests/tui/test_app.py
@@ -197,9 +197,12 @@ def mock_query_one(selector, *args):
         if isinstance(selector, type):
             name = selector.__name__
         else:
-            if "," in selector or "#" in selector:
+            if selector == "#input" or selector == "#input, InputArea":
                 return mock_input_area
-            return mock_footer
+            elif selector == "#status-bar" or selector == "#status-bar, StatusBar":
+                return mock_status_bar
+            name = "MainFooter"  # Default fallback
+
         mapping = {
             "MainFooter": mock_footer,
             "StatusBar": mock_status_bar,
@@ -217,9 +220,6 @@ def mock_query_one(selector, *args):
     tui_instance.worker = MagicMock()
     tui_instance.worker.coder = mock_coder
 
-    # Stub status_bar reference
-    tui_instance.status_bar = mock_status_bar
-
     # Mock AgentService - unknown UUID should return None (no prefix)
     monkeypatch.setattr(
         "cecli.helpers.agents.service.AgentService.get_instance",
@@ -239,3 +239,40 @@ def mock_query_one(selector, *args):
         timeout=5,
         agent_name=None,
     )
+
+
+def test_show_error_uses_query_one(tui_instance):
+    """
+    Test that show_error uses query_one to get the status bar and show a notification.
+    """
+    mock_status_bar = MagicMock()
+    tui_instance.query_one = MagicMock(return_value=mock_status_bar)
+
+    # Import StatusBar for the assertion
+    from cecli.tui.widgets import StatusBar
+
+    tui_instance.show_error("A test error", agent_name="test_agent")
+
+    # Assert query_one was called correctly
+    tui_instance.query_one.assert_called_once_with("#status-bar", StatusBar)
+
+    # Assert show_notification was called on the result of query_one
+    mock_status_bar.show_notification.assert_called_once_with(
+        "A test error",
+        severity="error",
+        timeout=5,
+        agent_name="test_agent",
+    )
+    # Test: error message for unknown agent should have agent_name=None
+    msg = {
+        "type": "error",
+        "message": "Something went wrong!",
+        "coder_uuid": "unknown_uuid",
+    }
+    tui_instance.handle_output_message(msg)
+    mock_status_bar.show_notification.assert_called_once_with(
+        "Something went wrong!",
+        severity="error",
+        timeout=5,
+        agent_name=None,
+    )