fix(gateway,windows): preserve restart watcher env

gateway/run.py

@@ -32,6 +32,7 @@ import logging
 import os
 import re
 import shlex
+import site
 import sys
 import signal
 import tempfile
@@ -135,6 +136,60 @@ _GATEWAY_SECRET_PATTERNS = (
 )
 
 
+def _ensure_windows_gateway_venv_imports() -> None:
+    """Make detached Windows gateway runs see the Hermes venv packages.
+
+    Some Windows restart paths run the gateway under uv's base ``pythonw.exe``
+    to avoid the venv launcher respawning a visible console interpreter.  That
+    mode can import the source tree via cwd/PYTHONPATH but still miss optional
+    packages installed only in ``venv/Lib/site-packages`` (notably the MCP SDK).
+    Patch the live process before MCP discovery so tool injection does not
+    depend on every launcher preserving PYTHONPATH perfectly.
+    """
+    if sys.platform != "win32":
+        return
+
+    project_root = Path(__file__).resolve().parent.parent
+    candidates: list[Path] = []
+    if os.environ.get("VIRTUAL_ENV"):
+        candidates.append(Path(os.environ["VIRTUAL_ENV"]))
+    candidates.append(project_root / "venv")
+
+    seen: set[str] = set()
+    for venv_dir in candidates:
+        try:
+            resolved_venv = venv_dir.resolve()
+        except OSError:
+            resolved_venv = venv_dir
+        venv_key = str(resolved_venv).lower()
+        if venv_key in seen:
+            continue
+        seen.add(venv_key)
+
+        site_packages = resolved_venv / "Lib" / "site-packages"
+        if not site_packages.exists():
+            continue
+
+        project_entry = str(project_root)
+        site_entry = str(site_packages)
+        if project_entry not in sys.path:
+            sys.path.insert(0, project_entry)
+        # addsitepackages() semantics matter here: pywin32, used by the MCP
+        # SDK on Windows, relies on .pth processing to expose pywintypes.
+        site.addsitedir(site_entry)
+        if site_entry in sys.path:
+            sys.path.remove(site_entry)
+        insert_at = 1 if sys.path and sys.path[0] == project_entry else 0
+        sys.path.insert(insert_at, site_entry)
+
+        os.environ["VIRTUAL_ENV"] = str(resolved_venv)
+        pythonpath = [project_entry, site_entry]
+        if os.environ.get("PYTHONPATH"):
+            pythonpath.append(os.environ["PYTHONPATH"])
+        os.environ["PYTHONPATH"] = os.pathsep.join(dict.fromkeys(pythonpath))
+        return
+
+
 def _gateway_platform_value(platform: Any) -> str:
     """Return a normalized gateway platform value for enums or raw strings."""
     return str(getattr(platform, "value", platform) or "").strip().lower()
@@ -4255,10 +4310,25 @@ class GatewayRunner(GatewayAuthorizationMixin, GatewayKanbanWatchersMixin, Gatew
                 )
                 """
             ).strip()
+            watcher_env = os.environ.copy()
+            # This watcher is intentionally outside the running gateway. If it
+            # inherits the gateway marker, `hermes gateway restart` refuses to
+            # run as a self-restart loop guard and the gateway stays stopped.
+            watcher_env.pop("_HERMES_GATEWAY", None)
+            project_root = Path(__file__).resolve().parent.parent
+            venv_dir = Path(watcher_env.get("VIRTUAL_ENV") or project_root / "venv")
+            site_packages = venv_dir / "Lib" / "site-packages"
+            if site_packages.exists():
+                watcher_env["VIRTUAL_ENV"] = str(venv_dir)
+                pythonpath = [str(project_root), str(site_packages)]
+                if watcher_env.get("PYTHONPATH"):
+                    pythonpath.append(watcher_env["PYTHONPATH"])
+                watcher_env["PYTHONPATH"] = os.pathsep.join(dict.fromkeys(pythonpath))
             subprocess.Popen(
                 [sys.executable, "-c", watcher, str(current_pid), *cmd_argv],
                 stdout=subprocess.DEVNULL,
                 stderr=subprocess.DEVNULL,
+                env=watcher_env,
                 **windows_detach_popen_kwargs(),
             )
             return
@@ -15910,6 +15980,8 @@ async def start_gateway(config: Optional[GatewayConfig] = None, replace: bool =
     atexit.register(remove_pid_file)
     atexit.register(release_gateway_runtime_lock)
 
+    _ensure_windows_gateway_venv_imports()
+
     # MCP tool discovery — run in an executor so the asyncio event loop
     # stays responsive even when a configured MCP server is slow or
     # unreachable.  discover_mcp_tools() uses a blocking 120s wait

tests/gateway/test_restart_drain.py

@@ -197,6 +197,7 @@ async def test_launch_detached_restart_command_uses_setsid(monkeypatch):
     runner, _adapter = make_restart_runner()
     popen_calls = []
 
+    monkeypatch.setattr(gateway_run.sys, "platform", "linux")
     monkeypatch.setattr(gateway_run, "_resolve_hermes_bin", lambda: ["/usr/bin/hermes"])
     monkeypatch.setattr(gateway_run.os, "getpid", lambda: 321)
     monkeypatch.setattr(shutil, "which", lambda cmd: "/usr/bin/setsid" if cmd == "setsid" else None)
@@ -219,6 +220,69 @@ async def test_launch_detached_restart_command_uses_setsid(monkeypatch):
     assert kwargs["stderr"] is subprocess.DEVNULL
 
 
+def test_windows_gateway_venv_imports_add_site_packages(monkeypatch, tmp_path):
+    venv_dir = tmp_path / "venv"
+    site_packages = venv_dir / "Lib" / "site-packages"
+    pth_extra = tmp_path / "pywin32_system32"
+    site_packages.mkdir(parents=True)
+    pth_extra.mkdir()
+    (site_packages / "pywin32.pth").write_text(str(pth_extra), encoding="utf-8")
+    project_root = str(gateway_run.Path(gateway_run.__file__).resolve().parent.parent)
+
+    monkeypatch.setattr(gateway_run.sys, "platform", "win32")
+    monkeypatch.setattr(gateway_run.sys, "path", ["existing"])
+    monkeypatch.setenv("VIRTUAL_ENV", str(venv_dir))
+    monkeypatch.setenv("PYTHONPATH", "already-there")
+
+    gateway_run._ensure_windows_gateway_venv_imports()
+
+    assert gateway_run.sys.path[:2] == [project_root, str(site_packages)]
+    assert str(pth_extra) in gateway_run.sys.path
+    assert gateway_run.os.environ["VIRTUAL_ENV"] == str(venv_dir.resolve())
+    pythonpath = gateway_run.os.environ["PYTHONPATH"].split(gateway_run.os.pathsep)
+    assert pythonpath[:3] == [project_root, str(site_packages), "already-there"]
+
+
+@pytest.mark.asyncio
+async def test_windows_detached_restart_scrubs_gateway_marker(monkeypatch, tmp_path):
+    runner, _adapter = make_restart_runner()
+    popen_calls = []
+    venv_dir = tmp_path / "venv"
+    site_packages = venv_dir / "Lib" / "site-packages"
+    site_packages.mkdir(parents=True)
+
+    monkeypatch.setattr(gateway_run.sys, "platform", "win32")
+    monkeypatch.setattr(gateway_run, "_resolve_hermes_bin", lambda: ["hermes"])
+    monkeypatch.setattr(gateway_run.os, "getpid", lambda: 321)
+    monkeypatch.setenv("_HERMES_GATEWAY", "1")
+    monkeypatch.setenv("VIRTUAL_ENV", str(venv_dir))
+
+    import hermes_cli._subprocess_compat as subprocess_compat
+
+    monkeypatch.setattr(
+        subprocess_compat,
+        "windows_detach_popen_kwargs",
+        lambda: {},
+    )
+
+    def fake_popen(cmd, **kwargs):
+        popen_calls.append((cmd, kwargs))
+        return MagicMock()
+
+    monkeypatch.setattr(subprocess, "Popen", fake_popen)
+
+    await runner._launch_detached_restart_command()
+
+    assert len(popen_calls) == 1
+    cmd, kwargs = popen_calls[0]
+    assert cmd[-3:] == ["hermes", "gateway", "restart"]
+    assert kwargs["env"].get("_HERMES_GATEWAY") is None
+    assert kwargs["env"]["VIRTUAL_ENV"] == str(venv_dir)
+    assert str(site_packages) in kwargs["env"]["PYTHONPATH"].split(gateway_run.os.pathsep)
+    assert kwargs["stdout"] is subprocess.DEVNULL
+    assert kwargs["stderr"] is subprocess.DEVNULL
+
+
 # ── Shutdown notification tests ──────────────────────────────────────