Fix user commands in terminal with function calling

openhands/agenthub/codeact_agent/codeact_agent.py

@@ -5,6 +5,7 @@
 from litellm import ModelResponse
 
 import openhands.agenthub.codeact_agent.function_calling as codeact_function_calling
+from openhands.agenthub.codeact_agent.function_calling import action_to_str
 from openhands.controller.agent import Agent
 from openhands.controller.state.state import State
 from openhands.core.config import AgentConfig
@@ -148,12 +149,14 @@ def get_action_message(
             action,
             (
                 AgentDelegateAction,
-                CmdRunAction,
                 IPythonRunCellAction,
                 FileEditAction,
                 BrowseInteractiveAction,
             ),
-        ) or (isinstance(action, AgentFinishAction) and action.source == 'agent'):
+        ) or (
+            isinstance(action, (AgentFinishAction, CmdRunAction))
+            and action.source == 'agent'
+        ):
             tool_metadata = action.tool_call_metadata
             assert tool_metadata is not None, (
                 'Tool call metadata should NOT be None when function calling is enabled. Action: '
@@ -184,6 +187,14 @@ def get_action_message(
                     content=content,
                 )
             ]
+        elif isinstance(action, CmdRunAction) and action.source == 'user':
+            content = [TextContent(text=action_to_str(action))]
+            return [
+                Message(
+                    role='user',
+                    content=content,
+                )
+            ]
         return []
 
     def get_observation_message(
@@ -219,6 +230,13 @@ def get_observation_message(
         message: Message
         max_message_chars = self.llm.config.max_message_chars
         if isinstance(obs, CmdOutputObservation):
+            # if the cause of this observation is a user action, show that clearly
+            if (
+                obs.cause
+                and isinstance(obs.cause, CmdRunAction)
+                and obs.cause.source == 'user'
+            ):
+                obs.content += f'\n[Observed result of command executed by user:\n{obs.cause.command}]'
             text = truncate_content(
                 obs.content + obs.interpreter_details, max_message_chars
             )

openhands/agenthub/codeact_agent/function_calling.py

@@ -521,3 +521,11 @@ def get_tools(
     else:
         tools.append(StrReplaceEditorTool)
     return tools
+
+
+def action_to_str(action: Action) -> str:
+    # for example, when the user runs a command from the terminal
+    # we want to show the command in the chat history
+    if isinstance(action, CmdRunAction) and action.source == 'user':
+        return f'User executed the command:\n{action.command}'
+    return ''