fix(gateway): /status command always shows 0 tokens

SessionEntry.total_tokens was initialized to 0 in the constructor but
never updated during the session lifecycle. The agent tracks cumulative
tokens on agent.session_total_tokens, but that value was never written
back to the session store.

Changes:
- session.py: update_session() accepts optional total_tokens parameter
- run.py: pipe agent.session_total_tokens through result dict into
  update_session() after each agent run (both success and error paths)
- test_status_command.py: updated assertion for new call signature

gateway/run.py

@@ -4068,6 +4068,7 @@ class GatewayRunner:
             self.session_store.update_session(
                 session_entry.session_key,
                 last_prompt_tokens=agent_result.get("last_prompt_tokens", 0),
+                total_tokens=agent_result.get("session_total_tokens", 0),
             )
 
             # Auto voice reply: send TTS audio before the text response
@@ -8712,11 +8713,14 @@ class GatewayRunner:
             _last_prompt_toks = 0
             _input_toks = 0
             _output_toks = 0
+            _session_total_toks = 0
             _agent = agent_holder[0]
             if _agent and hasattr(_agent, "context_compressor"):
                 _last_prompt_toks = getattr(_agent.context_compressor, "last_prompt_tokens", 0)
                 _input_toks = getattr(_agent, "session_prompt_tokens", 0)
                 _output_toks = getattr(_agent, "session_completion_tokens", 0)
+            if _agent:
+                _session_total_toks = getattr(_agent, "session_total_tokens", 0) or 0
             _resolved_model = getattr(_agent, "model", None) if _agent else None
 
             if not final_response:
@@ -8732,6 +8736,7 @@ class GatewayRunner:
                     "last_prompt_tokens": _last_prompt_toks,
                     "input_tokens": _input_toks,
                     "output_tokens": _output_toks,
+                    "session_total_tokens": _session_total_toks,
                     "model": _resolved_model,
                 }
             
@@ -8821,6 +8826,7 @@ class GatewayRunner:
                 "last_prompt_tokens": _last_prompt_toks,
                 "input_tokens": _input_toks,
                 "output_tokens": _output_toks,
+                "session_total_tokens": _session_total_toks,
                 "model": _resolved_model,
                 "session_id": effective_session_id,
                 "response_previewed": result.get("response_previewed", False),

gateway/session.py

@@ -770,6 +770,7 @@ class SessionStore:
         self,
         session_key: str,
         last_prompt_tokens: int = None,
+        total_tokens: int = None,
     ) -> None:
         """Update lightweight session metadata after an interaction."""
         with self._lock:
@@ -780,6 +781,8 @@ class SessionStore:
                 entry.updated_at = _now()
                 if last_prompt_tokens is not None:
                     entry.last_prompt_tokens = last_prompt_tokens
+                if total_tokens is not None:
+                    entry.total_tokens = total_tokens
                 self._save()
 
     def suspend_session(self, session_key: str) -> bool:

tests/gateway/test_status_command.py

@@ -150,6 +150,7 @@ async def test_handle_message_persists_agent_token_counts(monkeypatch):
     runner.session_store.update_session.assert_called_once_with(
         session_entry.session_key,
         last_prompt_tokens=80,
+        total_tokens=0,
     )