fix: sanitize messages and ensure 'content' for strict LLM providers

- Strip non-standard keys like 'reasoning_content' before sending to LLM
- Always include 'content' key in assistant messages (required by StepFun)
- Add _sanitize_messages to LiteLLMProvider to prevent 400 BadRequest errors

nanobot/agent/context.py

@@ -227,9 +227,9 @@ To recall past events, grep {workspace_path}/memory/HISTORY.md"""
         """
         msg: dict[str, Any] = {"role": "assistant"}
 
-        # Omit empty content — some backends reject empty text blocks
-        if content:
-            msg["content"] = content
+        # Always include content — some providers (e.g. StepFun) reject
+        # assistant messages that omit the key entirely.
+        msg["content"] = content
 
         if tool_calls:
             msg["tool_calls"] = tool_calls

nanobot/providers/litellm_provider.py

@@ -12,6 +12,12 @@ from nanobot.providers.base import LLMProvider, LLMResponse, ToolCallRequest
 from nanobot.providers.registry import find_by_model, find_gateway
 
 
+# Keys that are part of the OpenAI chat-completion message schema.
+# Anything else (e.g. reasoning_content, timestamp) is stripped before sending
+# to avoid "Unrecognized chat message" errors from strict providers like StepFun.
+_ALLOWED_MSG_KEYS = frozenset({"role", "content", "tool_calls", "tool_call_id", "name"})
+
+
 class LiteLLMProvider(LLMProvider):
     """
     LLM provider using LiteLLM for multi-provider support.
@@ -103,6 +109,24 @@ class LiteLLMProvider(LLMProvider):
                     kwargs.update(overrides)
                     return
     
+    @staticmethod
+    def _sanitize_messages(messages: list[dict[str, Any]]) -> list[dict[str, Any]]:
+        """Strip non-standard keys from messages for strict providers.
+
+        Some providers (e.g. StepFun via OpenRouter) reject messages that
+        contain extra keys like ``reasoning_content``. This method keeps
+        only the keys defined in the OpenAI chat-completion schema and
+        ensures every assistant message has a ``content`` key.
+        """
+        sanitized = []
+        for msg in messages:
+            clean = {k: v for k, v in msg.items() if k in _ALLOWED_MSG_KEYS}
+            # Strict providers require "content" even when assistant only has tool_calls
+            if clean.get("role") == "assistant" and "content" not in clean:
+                clean["content"] = None
+            sanitized.append(clean)
+        return sanitized
+
     async def chat(
         self,
         messages: list[dict[str, Any]],
@@ -132,7 +156,7 @@ class LiteLLMProvider(LLMProvider):
         
         kwargs: dict[str, Any] = {
             "model": model,
-            "messages": messages,
+            "messages": self._sanitize_messages(messages),
             "max_tokens": max_tokens,
             "temperature": temperature,
         }