diff --git a/metagpt/roles/role.py b/metagpt/roles/role.py
index e0f8a7ea6..b1773b739 100644
--- a/metagpt/roles/role.py
+++ b/metagpt/roles/role.py
@@ -455,8 +455,8 @@ class Role(SerializationMixin, ContextMixin, BaseModel):
         rsp = Message(content="No actions taken yet", cause_by=Action)  # will be overwritten after Role _act
         while actions_taken < self.rc.max_react_loop:
             # think
-            await self._think()
-            if self.rc.todo is None:
+            has_todo = await self._think()
+            if not has_todo:
                 break
             # act
             logger.debug(f"{self._setting}: {self.rc.state=}, will do {self.rc.todo}")
diff --git a/metagpt/strategy/thinking_command.py b/metagpt/strategy/thinking_command.py
index 19604ad39..7b48cae51 100644
--- a/metagpt/strategy/thinking_command.py
+++ b/metagpt/strategy/thinking_command.py
@@ -36,7 +36,7 @@ class Command(Enum):
     PUBLISH_MESSAGE = CommandDef(
         name="publish_message",
         signature="publish_message(content: str, send_to: str)",
-        desc="Publish a message to a team member, use member name to fill send_to args. You may copy the full original content or add additional information from upstream. This will make team members start their work. DONT omit any necessary info such as path, link, environment from original content to team members because you are their sole info source. However, if the original content is long or contains concrete info, you should forward the message using forward_message instead of publishing it.",
+        desc="Publish a message to a team member, use member name to fill send_to args. You may copy the full original content or add additional information from upstream. This will make team members start their work. DONT omit any necessary info such as path, link, environment from original content to team members because you are their sole info source.",
     )
     REPLY_TO_HUMAN = CommandDef(
         name="reply_to_human",