fix(gateway): read /status token totals from SessionDB (#17158)

/status was reading session_entry.total_tokens from the in-memory
SessionStore (gateway/session.py), which the agent never writes to —
so the token count was always 0.

The agent already persists token deltas to the SQLite SessionDB
(run_agent.py:11497) for every platform with a session_id. Route
/status through that single source of truth instead of duplicating
token writes into a second store.

Fix:
- gateway/run.py: _handle_status_command now calls
  self._session_db.get_session(session_id) and sums the five token
  component columns (input/output/cache_read/cache_write/reasoning).
  Falls back to 0 when no SessionDB is configured or no row exists.
- Two new regression tests covering the populated-row and
  missing-row paths.

Co-authored-by: Hermes <127238744+teknium1@users.noreply.github.com>

gateway/run.py

@@ -6568,11 +6568,30 @@ class GatewayRunner:
         queue_depth = self._queue_depth(session_key, adapter=adapter)
 
         title = None
+        # Pull token totals from the SQLite session DB rather than the
+        # in-memory SessionStore.  The agent's per-turn token deltas are
+        # persisted into sessions_db (run_agent.py), not into SessionEntry,
+        # so session_entry.total_tokens is always 0.  SessionDB is the
+        # single source of truth; reading it here keeps /status accurate
+        # without duplicating token writes into two stores.
+        db_total_tokens = 0
         if self._session_db:
             try:
                 title = self._session_db.get_session_title(session_entry.session_id)
             except Exception:
                 title = None
+            try:
+                row = self._session_db.get_session(session_entry.session_id)
+                if row:
+                    db_total_tokens = (
+                        (row.get("input_tokens") or 0)
+                        + (row.get("output_tokens") or 0)
+                        + (row.get("cache_read_tokens") or 0)
+                        + (row.get("cache_write_tokens") or 0)
+                        + (row.get("reasoning_tokens") or 0)
+                    )
+            except Exception:
+                db_total_tokens = 0
 
         lines = [
             "📊 **Hermes Gateway Status**",
@@ -6584,7 +6603,7 @@ class GatewayRunner:
         lines.extend([
             f"**Created:** {session_entry.created_at.strftime('%Y-%m-%d %H:%M')}",
             f"**Last Activity:** {session_entry.updated_at.strftime('%Y-%m-%d %H:%M')}",
-            f"**Tokens:** {session_entry.total_tokens:,}",
+            f"**Tokens:** {db_total_tokens:,}",
             f"**Agent Running:** {'Yes ⚡' if is_running else 'No'}",
         ])
         if queue_depth:

tests/gateway/test_status_command.py

@@ -55,6 +55,9 @@ def _make_runner(session_entry: SessionEntry, *, platform: Platform = Platform.T
     runner._pending_approvals = {}
     runner._session_db = MagicMock()
     runner._session_db.get_session_title.return_value = None
+    # Default: no DB row → /status reports 0 tokens.  Tests that exercise
+    # the populated path override this.
+    runner._session_db.get_session.return_value = None
     runner._reasoning_config = None
     runner._provider_routing = {}
     runner._fallback_model = None
@@ -80,6 +83,14 @@ async def test_status_command_reports_running_agent_without_interrupt(monkeypatc
         total_tokens=321,
     )
     runner = _make_runner(session_entry)
+    # Token total comes from the SQLite SessionDB, not SessionEntry.
+    runner._session_db.get_session.return_value = {
+        "input_tokens": 200,
+        "output_tokens": 121,
+        "cache_read_tokens": 0,
+        "cache_write_tokens": 0,
+        "reasoning_tokens": 0,
+    }
     running_agent = MagicMock()
     runner._running_agents[build_session_key(_make_source())] = running_agent
 
@@ -113,6 +124,56 @@ async def test_status_command_includes_session_title_when_present():
     assert "**Title:** My titled session" in result
 
 
+@pytest.mark.asyncio
+async def test_status_command_reads_token_totals_from_session_db():
+    """Regression test for #17158: /status must source token totals from the
+    SQLite SessionDB (where run_agent.py persists them) and sum all component
+    counts, not from SessionEntry (which the agent never writes)."""
+    session_entry = SessionEntry(
+        session_key=build_session_key(_make_source()),
+        session_id="sess-1",
+        created_at=datetime.now(),
+        updated_at=datetime.now(),
+        platform=Platform.TELEGRAM,
+        chat_type="dm",
+        total_tokens=0,  # SessionEntry never gets written to — always 0.
+    )
+    runner = _make_runner(session_entry)
+    runner._session_db.get_session.return_value = {
+        "input_tokens": 1000,
+        "output_tokens": 250,
+        "cache_read_tokens": 500,
+        "cache_write_tokens": 100,
+        "reasoning_tokens": 50,
+    }
+
+    result = await runner._handle_message(_make_event("/status"))
+
+    # 1000 + 250 + 500 + 100 + 50 = 1,900
+    assert "**Tokens:** 1,900" in result
+
+
+@pytest.mark.asyncio
+async def test_status_command_tokens_zero_when_session_db_row_missing():
+    """When the SessionDB has no row for the current session yet (fresh
+    session, no agent calls), /status reports 0 without raising."""
+    session_entry = SessionEntry(
+        session_key=build_session_key(_make_source()),
+        session_id="sess-1",
+        created_at=datetime.now(),
+        updated_at=datetime.now(),
+        platform=Platform.TELEGRAM,
+        chat_type="dm",
+        total_tokens=999,  # This should be ignored.
+    )
+    runner = _make_runner(session_entry)
+    runner._session_db.get_session.return_value = None
+
+    result = await runner._handle_message(_make_event("/status"))
+
+    assert "**Tokens:** 0" in result
+
+
 @pytest.mark.asyncio
 async def test_agents_command_reports_active_agents_and_processes(monkeypatch):
     session_key = build_session_key(_make_source())