(Hotfix): Track reason for Error AgentState

frontend/__tests__/components/chat/action-suggestions.test.tsx

@@ -22,11 +22,11 @@ vi.mock("#/context/auth-context", () => ({
 describe("ActionSuggestions", () => {
   // Setup mocks for each test
   vi.clearAllMocks();
-  
+
   (useAuth as any).mockReturnValue({
     githubTokenIsSet: true,
   });
-  
+
   (useSelector as any).mockReturnValue({
     selectedRepository: "test-repo",
   });
@@ -66,16 +66,16 @@ describe("ActionSuggestions", () => {
   it("should have different prompts for 'Push to Branch' and 'Push & Create PR' buttons", () => {
     // This test verifies that the prompts are different in the component
     const component = render(<ActionSuggestions onSuggestionsClick={() => {}} />);
-    
+
     // Get the component instance to access the internal values
     const pushBranchPrompt = "Please push the changes to a remote branch on GitHub, but do NOT create a pull request. Please use the exact SAME branch name as the one you are currently on.";
     const createPRPrompt = "Please push the changes to GitHub and open a pull request. Please create a meaningful branch name that describes the changes.";
-    
+
     // Verify the prompts are different
     expect(pushBranchPrompt).not.toEqual(createPRPrompt);
-    
+
     // Verify the PR prompt mentions creating a meaningful branch name
     expect(createPRPrompt).toContain("meaningful branch name");
     expect(createPRPrompt).not.toContain("SAME branch name");
   });
-});
+});

frontend/src/i18n/translation.json

@@ -2078,6 +2078,7 @@
         "tr": "Ajan hız sınırına ulaştı",
         "ja": "エージェントがレート制限中"
     },
+
     "CHAT_INTERFACE$AGENT_PAUSED_MESSAGE": {
         "en": "Agent has paused.",
         "de": "Agent pausiert.",

openhands/controller/agent_controller.py

@@ -228,11 +228,14 @@ async def _react_to_exception(
         e: Exception,
     ):
         """React to an exception by setting the agent state to error and sending a status message."""
-        await self.set_agent_state_to(AgentState.ERROR)
+        # Store the error reason before setting the agent state
+        self.state.last_error = f'{type(e).__name__}: {str(e)}'
+
         if self.status_callback is not None:
             err_id = ''
             if isinstance(e, AuthenticationError):
                 err_id = 'STATUS$ERROR_LLM_AUTHENTICATION'
+                self.state.last_error = 'LLM authentication error'
             elif isinstance(
                 e,
                 (
@@ -242,14 +245,24 @@ async def _react_to_exception(
                 ),
             ):
                 err_id = 'STATUS$ERROR_LLM_SERVICE_UNAVAILABLE'
+                self.state.last_error = 'LLM service is unavailable'
             elif isinstance(e, InternalServerError):
                 err_id = 'STATUS$ERROR_LLM_INTERNAL_SERVER_ERROR'
+                self.state.last_error = 'LLM internal server error'
             elif isinstance(e, BadRequestError) and 'ExceededBudget' in str(e):
                 err_id = 'STATUS$ERROR_LLM_OUT_OF_CREDITS'
+                # Set error reason for budget exceeded
+                self.state.last_error = 'budget exceeded (out of credits)'
+                # Use ERROR state with reason instead of separate state
+                await self.set_agent_state_to(AgentState.ERROR)
+                return
             elif isinstance(e, RateLimitError):
                 await self.set_agent_state_to(AgentState.RATE_LIMITED)
                 return
-            self.status_callback('error', err_id, type(e).__name__ + ': ' + str(e))
+            self.status_callback('error', err_id, self.state.last_error)
+
+        # Set the agent state to ERROR after storing the reason
+        await self.set_agent_state_to(AgentState.ERROR)
 
     def step(self):
         asyncio.create_task(self._step_with_exception_handling())
@@ -582,8 +595,14 @@ async def set_agent_state_to(self, new_state: AgentState) -> None:
             self.event_stream.add_event(self._pending_action, EventSource.AGENT)
 
         self.state.agent_state = new_state
+
+        # Create observation with reason field if it's an error state
+        reason = ''
+        if new_state == AgentState.ERROR:
+            reason = self.state.last_error
+
         self.event_stream.add_event(
-            AgentStateChangedObservation('', self.state.agent_state),
+            AgentStateChangedObservation('', self.state.agent_state, reason),
             EventSource.ENVIRONMENT,
         )
 

openhands/events/observation/agent.py

@@ -10,6 +10,7 @@ class AgentStateChangedObservation(Observation):
     """This data class represents the result from delegating to another agent"""
 
     agent_state: str
+    reason: str = ''
     observation: str = ObservationType.AGENT_STATE_CHANGED
 
     @property
@@ -113,7 +114,7 @@ def __str__(self) -> str:
                     f'repo_instructions={self.repo_instructions[:20]}...',
                     f'runtime_hosts={self.runtime_hosts}',
                     f'additional_agent_instructions={self.additional_agent_instructions[:20]}...',
-                    f'date={self.date}'
+                    f'date={self.date}',
                 ]
             )
         else:

openhands/runtime/plugins/vscode/settings.json

@@ -1,4 +1,4 @@
 {
     "workbench.colorTheme": "Default Dark Modern",
     "workbench.startupEditor": "none"
-}
+}

tests/unit/test_agent_controller.py

@@ -17,6 +17,7 @@
 from openhands.events.action.agent import RecallAction
 from openhands.events.event import RecallType
 from openhands.events.observation import (
+    AgentStateChangedObservation,
     ErrorObservation,
 )
 from openhands.events.observation.agent import RecallObservation
@@ -216,9 +217,17 @@ def on_event_memory(event: Event):
     print(f'state: {state}')
     events = list(test_event_stream.get_events())
     print(f'event_stream: {events}')
+    error_observations = test_event_stream.get_matching_events(
+        reverse=True, limit=1, event_types=(AgentStateChangedObservation)
+    )
+    assert len(error_observations) == 1
+    error_observation = error_observations[0]
     assert state.iteration == 3
     assert state.agent_state == AgentState.ERROR
     assert state.last_error == 'AgentStuckInLoopError: Agent got stuck in a loop'
+    assert (
+        error_observation.reason == 'AgentStuckInLoopError: Agent got stuck in a loop'
+    )
     assert len(events) == 11
 
 
@@ -621,6 +630,17 @@ def on_event_memory(event: Event):
         state.last_error
         == 'RuntimeError: Agent reached maximum iteration in headless mode. Current iteration: 3, max iteration: 3'
     )
+    error_observations = test_event_stream.get_matching_events(
+        reverse=True, limit=1, event_types=(AgentStateChangedObservation)
+    )
+    assert len(error_observations) == 1
+    error_observation = error_observations[0]
+
+    assert (
+        error_observation.reason
+        == 'RuntimeError: Agent reached maximum iteration in headless mode. Current iteration: 3, max iteration: 3'
+    )
+
     assert (
         state.metrics.accumulated_cost == 10.0 * 3
     ), f'Expected accumulated cost to be 30.0, but got {state.metrics.accumulated_cost}'
@@ -837,6 +857,16 @@ def on_event_memory(event: Event):
         == 'LLMContextWindowExceedError: Conversation history longer than LLM context window limit. Consider turning on enable_history_truncation config to avoid this error'
     )
 
+    error_observations = test_event_stream.get_matching_events(
+        reverse=True, limit=1, event_types=(AgentStateChangedObservation)
+    )
+    assert len(error_observations) == 1
+    error_observation = error_observations[0]
+    assert (
+        error_observation.reason
+        == 'LLMContextWindowExceedError: Conversation history longer than LLM context window limit. Consider turning on enable_history_truncation config to avoid this error'
+    )
+
     # Check that the context window exceeded error was raised during the run
     assert step_state.has_errored