fix(dashboard): reduce idle disk writes

hermes_cli/main.py

@@ -10325,25 +10325,10 @@ def cmd_dashboard(args):
         # the missing-provider state if it matters.
         print(f"⚠ Plugin discovery failed: {exc}", file=sys.stderr)
 
-    # Desktop chat uses the dashboard's in-process /api/ws gateway, which builds
-    # agents via tui_gateway.server._make_agent.  That path only snapshots the
-    # tool registry — it never starts MCP discovery (the stdio TUI does that in
-    # tui_gateway/entry.py, which the dashboard process doesn't run).  Without
-    # this, a profile's configured MCP servers never connect, so desktop
-    # sessions show no MCP tools.  Spawn discovery in the background here so a
-    # slow/dead server can't block dashboard startup.
-    try:
-        from hermes_cli.mcp_startup import start_background_mcp_discovery
-
-        start_background_mcp_discovery(
-            logger=logger,
-            thread_name="dashboard-mcp-discovery",
-        )
-    except Exception:
-        logger.debug(
-            "Background MCP tool discovery failed at dashboard startup",
-            exc_info=True,
-        )
+    # Keep an unvisited dashboard disk-idle: MCP discovery may launch stdio
+    # servers and their stderr log streams. The dashboard chat WebSocket starts
+    # discovery on first use before agent construction, preserving MCP tools
+    # without running the MCP stack in an idle management server.
 
     from hermes_cli.web_server import start_server
 

hermes_state.py

@@ -2509,8 +2509,10 @@ def close(self):
         """Close the database connection.
 
         Drains queued token deltas first (the background writer needs the
-        connection), then attempts a TRUNCATE WAL checkpoint so that
-        exiting processes help shrink the WAL file.
+        connection). Writable connections then attempt a TRUNCATE WAL
+        checkpoint so that exiting processes help shrink the WAL file.
+        Read-only probes close without checkpointing, preserving their
+        no-write contract.
         """
         self._stop_token_writer()
         # The atexit hook holds a strong reference to this instance (bound
@@ -2538,10 +2540,13 @@ def close(self):
         self._read_local.conn = None
         with self._lock:
             if self._conn:
-                try:
-                    self._conn.execute("PRAGMA wal_checkpoint(TRUNCATE)")
-                except Exception as exc:
-                    logger.debug("WAL checkpoint (TRUNCATE) at close failed: %s", exc)
+                if not self.read_only:
+                    try:
+                        self._conn.execute("PRAGMA wal_checkpoint(TRUNCATE)")
+                    except Exception as exc:
+                        logger.debug(
+                            "WAL checkpoint (TRUNCATE) at close failed: %s", exc
+                        )
                 self._conn.close()
                 self._conn = None
 

tests/hermes_cli/test_dashboard_unified_launch.py

@@ -84,6 +84,42 @@ def test_desktop_profile_backend_skips_machine_dashboard_reroute(self, main_mod,
         assert listening_calls == []
         assert execs == []
 
+    def test_dashboard_defers_mcp_discovery_until_ws_backend(self, main_mod, monkeypatch):
+        """An unvisited dashboard must not launch MCP servers at startup.
 
+        The /api/ws sidecar starts discovery on first chat connection, before
+        agent construction, so MCP tools are still available without keeping an
+        idle dashboard writing mcp-stderr.log.
+        """
+        monkeypatch.setattr(
+            "hermes_cli.profiles.get_active_profile_name", lambda: "default"
+        )
+        monkeypatch.delenv("HERMES_WEB_DIST", raising=False)
+        monkeypatch.setattr(main_mod, "_sync_bundled_skills_quietly", lambda: None)
+        monkeypatch.setattr(main_mod, "_build_web_ui", lambda *_a, **_k: True)
+        monkeypatch.setitem(sys.modules, "fastapi", types.SimpleNamespace())
+        monkeypatch.setitem(sys.modules, "uvicorn", types.SimpleNamespace())
+        monkeypatch.setitem(
+            sys.modules,
+            "hermes_logging",
+            types.SimpleNamespace(setup_logging=lambda **_k: None),
+        )
+        monkeypatch.setitem(
+            sys.modules,
+            "hermes_cli.plugins",
+            types.SimpleNamespace(discover_plugins=lambda: None),
+        )
+        calls = []
+        monkeypatch.setattr(
+            "hermes_cli.mcp_startup.start_background_mcp_discovery",
+            lambda **kwargs: calls.append(kwargs),
+        )
+        monkeypatch.setitem(
+            sys.modules,
+            "hermes_cli.web_server",
+            types.SimpleNamespace(start_server=lambda **_kwargs: None),
+        )
 
+        main_mod.cmd_dashboard(_args())
 
+        assert calls == []

tests/hermes_cli/test_web_server.py

@@ -3517,4 +3517,3 @@ async def get(self, *args, **kwargs):
         assert self.ws.DASHBOARD_HEALTH.selftest_http_status == 500
         assert self.ws.DASHBOARD_HEALTH.snapshot()["status"] == "degraded"
 
-

tests/test_hermes_state.py

@@ -2135,6 +2135,33 @@ def _assume_fixed_sqlite(self, monkeypatch):
             hermes_state, "is_sqlite_wal_reset_vulnerable", lambda version_info=None: False
         )
 
+    def test_read_only_close_does_not_checkpoint(self, tmp_path, monkeypatch):
+        """Read-only dashboard probes must not trigger WAL checkpoints on close."""
+
+        class _FakeConn:
+            def __init__(self):
+                self.executed = []
+                self.closed = False
+                self.row_factory = None
+
+            def execute(self, sql, params=()):
+                self.executed.append(sql)
+                return self
+
+            def fetchone(self):
+                return None
+
+            def close(self):
+                self.closed = True
+
+        conn = _FakeConn()
+        monkeypatch.setattr(hermes_state.sqlite3, "connect", lambda *a, **kw: conn)
+
+        session_db = SessionDB(db_path=tmp_path / "state.db", read_only=True)
+        session_db.close()
+
+        assert conn.closed is True
+        assert not any("wal_checkpoint" in sql for sql in conn.executed)
 
     def test_sets_wal_on_fresh_connection(self, tmp_path):
         """Probe sees 'delete', then set-pragma runs and returns 'wal'."""