fix(classic): convert mid-conversation system messages to user messages

Some LLM providers (notably Anthropic) don't support system messages
in the middle of a conversation. Changed ChatMessage.system() to
ChatMessage.user() for all mid-conversation context messages across
components (action history, context, skills, system clock, todo,
error reporting, LATS, and multi-agent debate strategies).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

classic/forge/forge/components/action_history/action_history.py

@@ -93,7 +93,7 @@ class ActionHistoryComponent(
 
         if step_summaries:
             step_summaries_fmt = "\n\n".join(step_summaries)
-            yield ChatMessage.system(
+            yield ChatMessage.user(
                 f"## Progress on your Task so far\n"
                 "Here is a summary of the steps that you have executed so far, "
                 "use this as your consideration for determining the next action!\n"

classic/forge/forge/components/context/context.py

@@ -50,7 +50,7 @@ class ContextComponent(MessageProvider, CommandProvider):
 
     def get_messages(self) -> Iterator[ChatMessage]:
         if self.context:
-            yield ChatMessage.system(
+            yield ChatMessage.user(
                 "## Context\n"
                 f"{self.context.format_numbered(self.workspace)}\n\n"
                 "When a context item is no longer needed and you are not done yet, "

classic/forge/forge/components/skills/skill_component.py

@@ -99,7 +99,7 @@ class SkillComponent(
                 catalog_lines.append(
                     f"- **{name}**{loaded_marker}: {skill.metadata.description}"
                 )
-            yield ChatMessage.system("\n".join(catalog_lines))
+            yield ChatMessage.user("\n".join(catalog_lines))
 
         # Provide loaded skill content
         for name, skill in self._loaded_skills.items():
@@ -117,7 +117,7 @@ class SkillComponent(
                         loaded = " [loaded]" if f in skill.additional_files else ""
                         skill_content.append(f"- `{f}`{loaded}")
 
-                yield ChatMessage.system("\n".join(skill_content))
+                yield ChatMessage.user("\n".join(skill_content))
 
     # -------------------------------------------------------------------------
     # CommandProvider methods

classic/forge/forge/components/system/system.py

@@ -85,7 +85,7 @@ class SystemComponent(DirectiveProvider, MessageProvider, CommandProvider):
 
     def get_messages(self) -> Iterator[ChatMessage]:
         # Clock
-        yield ChatMessage.system(
+        yield ChatMessage.user(
             f"## Clock\nThe current time and date is {time.strftime('%c')}"
         )
 

classic/forge/forge/components/todo/todo.py

@@ -208,7 +208,7 @@ class TodoComponent(
         if completed:
             lines.append(f"\n**Completed:** {len(completed)} task(s)")
 
-        yield ChatMessage.system("\n".join(lines))
+        yield ChatMessage.user("\n".join(lines))
 
     # -------------------------------------------------------------------------
     # Helper Methods

classic/original_autogpt/autogpt/agents/agent.py

@@ -342,7 +342,7 @@ class Agent(BaseAgent[AnyActionProposal], Configurable[AgentSettings]):
         self, prompt: ChatPrompt, exception: Optional[Exception] = None
     ) -> AnyActionProposal:
         if exception:
-            prompt.messages.append(ChatMessage.system(f"Error: {exception}"))
+            prompt.messages.append(ChatMessage.user(f"Error: {exception}"))
 
         # Build thinking/reasoning kwargs from app config
         thinking_kwargs: dict[str, Any] = {}

classic/original_autogpt/autogpt/agents/prompt_strategies/lats.py

@@ -235,7 +235,7 @@ class LATSPromptStrategy(BaseMultiStepPromptStrategy):
                 ChatMessage.system(system_prompt),
                 ChatMessage.user(f'Task: """{task}"""'),
                 *messages,
-                ChatMessage.system(lats_context),
+                ChatMessage.user(lats_context),
                 ChatMessage.user(self._get_phase_instruction()),
             ],
             prefill_response='{\n    "thoughts":',

classic/original_autogpt/autogpt/agents/prompt_strategies/multi_agent_debate.py

@@ -219,7 +219,7 @@ class MultiAgentDebateStrategy(BaseMultiStepPromptStrategy):
                 ChatMessage.system(system_prompt),
                 ChatMessage.user(f'Task: """{task}"""'),
                 *messages,
-                ChatMessage.system(debate_context),
+                ChatMessage.user(debate_context),
                 ChatMessage.user(self._get_phase_instruction(task)),
             ],
             prefill_response='{\n    "thoughts":',