fix(mcp): allow background discovery retry after a run that connected nothing

start_background_mcp_discovery() sets _mcp_discovery_started once and never
resets it. If the first background run exits without connecting any MCP
server (startup cancellation, OOM restart, transient network failure), every
later call returns immediately and the process is permanently stuck with
zero MCP tools until a full restart.

Fix: when discovery is marked started but the thread is dead and no server
is connected, reset the flag and spawn a fresh discovery thread. Also log a
WARNING when a discovery run completes with zero connected servers, so the
condition is visible instead of silent.

Caught in production on a long-running gateway fleet where a gateway
restarted under memory pressure and came back with all MCP tools missing.

tests/hermes_cli/test_mcp_startup.py

@@ -222,3 +222,76 @@ def test_init_agent_waits_for_mcp_discovery_before_agent_build(monkeypatch):
     monkeypatch.setattr(cli_mod, "AIAgent", _fake_agent)
 
     assert cli._init_agent() is True
+
+
+def _retry_logger():
+    return types.SimpleNamespace(
+        debug=lambda *_a, **_k: None,
+        warning=lambda *_a, **_k: None,
+    )
+
+
+def _install_retry_stubs(monkeypatch, *, connected: bool, calls: dict):
+    monkeypatch.setitem(
+        sys.modules,
+        "hermes_cli.config",
+        types.SimpleNamespace(
+            read_raw_config=lambda: {"mcp_servers": {"demo": {"transport": "stdio"}}},
+        ),
+    )
+    monkeypatch.setitem(
+        sys.modules,
+        "tools.mcp_oauth",
+        types.SimpleNamespace(suppress_interactive_oauth=lambda: nullcontext()),
+    )
+    monkeypatch.setitem(
+        sys.modules,
+        "tools.mcp_tool",
+        types.SimpleNamespace(
+            discover_mcp_tools=lambda: calls.__setitem__("mcp", calls["mcp"] + 1),
+            get_mcp_status=lambda: [{"connected": connected}],
+        ),
+    )
+
+
+def test_background_discovery_retries_after_dead_thread_with_zero_connected(monkeypatch):
+    """A finished discovery run that connected nothing must not pin the
+    process in a 'discovery already started' state: the next call should be
+    allowed to retry (e.g. after startup cancellation or an OOM restart)."""
+    calls = {"mcp": 0}
+    _install_retry_stubs(monkeypatch, connected=False, calls=calls)
+
+    mcp_startup.start_background_mcp_discovery(
+        logger=_retry_logger(), thread_name="test-mcp-retry-1"
+    )
+    thread = mcp_startup._mcp_discovery_thread
+    if thread is not None:
+        thread.join(timeout=1.0)
+    assert calls["mcp"] == 1
+
+    mcp_startup.start_background_mcp_discovery(
+        logger=_retry_logger(), thread_name="test-mcp-retry-2"
+    )
+    thread = mcp_startup._mcp_discovery_thread
+    if thread is not None:
+        thread.join(timeout=1.0)
+    assert calls["mcp"] == 2
+
+
+def test_background_discovery_does_not_retry_when_servers_connected(monkeypatch):
+    """Once at least one MCP server is connected, repeat calls stay no-ops."""
+    calls = {"mcp": 0}
+    _install_retry_stubs(monkeypatch, connected=True, calls=calls)
+
+    mcp_startup.start_background_mcp_discovery(
+        logger=_retry_logger(), thread_name="test-mcp-noretry-1"
+    )
+    thread = mcp_startup._mcp_discovery_thread
+    if thread is not None:
+        thread.join(timeout=1.0)
+    assert calls["mcp"] == 1
+
+    mcp_startup.start_background_mcp_discovery(
+        logger=_retry_logger(), thread_name="test-mcp-noretry-2"
+    )
+    assert calls["mcp"] == 1