From 308d12e1fdf622520bb2c70664f6ad2c3fedd66e Mon Sep 17 00:00:00 2001
From: Teknium <127238744+teknium1@users.noreply.github.com>
Date: Wed, 9 Sep 2026 12:21:03 -0700
Subject: [PATCH] fix(tests): live-system guard lets a gateway start inside a
 container exec

#106623 (ca16cafee40) blocks any test subprocess whose argv resolves to
`hermes gateway run|start|restart`, so the harness cannot spawn a runtime that
outlives the worker and restarts the developer's gateway. The matcher strips the
`-p <profile>` selector and reads through `sh -c`, so it also fired on
`docker exec -u hermes <ctr> sh -c 'hermes -p x gateway start'` and broke
tests/docker/test_profile_gateway.py on every Docker build since. A gateway
launched inside a container cannot reach the host unit or webhook port; the
guard now skips commands whose argv[0] is a container runtime (docker, podman,
nerdctl). Host-side spawns stay blocked.
---
 tests/conftest.py               | 17 ++++++++++++++++-
 tests/test_live_system_guard.py | 26 ++++++++++++++++++++++++++
 2 files changed, 42 insertions(+), 1 deletion(-)

diff --git a/tests/conftest.py b/tests/conftest.py
index bac6334869..acaca8d8c5 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -1428,6 +1428,14 @@ def _live_system_guard(request, monkeypatch):
         "daemon-reload", "try-restart", "reload-or-restart",
     )
     _PROCESS_KILLERS = ("pkill", "killall", "taskkill", "skill", "fuser")
+    _CONTAINER_RUNTIMES = ("docker", "podman", "nerdctl")
+
+    def _first_token_basename(cmd_str: str) -> str:
+        try:
+            tokens = _shlex.split(cmd_str)
+        except ValueError:
+            tokens = cmd_str.split()
+        return tokens[0].rsplit("/", 1)[-1].lower() if tokens else ""
     # Shell/launcher executables whose arguments are themselves commands —
     # argv[0]-only scanning must not exempt what they wrap.
     _WRAPPER_COMMANDS = (
@@ -1563,7 +1571,14 @@ def _live_system_guard(request, monkeypatch):
         # sibling refactor moved the spawn seam and left tests patching the
         # facade. The canonical matcher, never an argv substring.
         from gateway.status import _gateway_command_subcommand
-        if not lookalike_ok and _gateway_command_subcommand(cmd_str) in ("run", "start", "restart"):
+        # A gateway launched INSIDE a container (`docker exec … hermes gateway start`) cannot
+        # reach the host's systemd unit or webhook port; tests/docker/ exists to exercise it.
+        in_container = _first_token_basename(cmd_str) in _CONTAINER_RUNTIMES
+        if (
+            not lookalike_ok
+            and not in_container
+            and _gateway_command_subcommand(cmd_str) in ("run", "start", "restart")
+        ):
             raise RuntimeError(
                 f"tests/conftest.py live-system guard: blocked "
                 f"subprocess.{name}({cmd!r}) — this would spawn a REAL "
diff --git a/tests/test_live_system_guard.py b/tests/test_live_system_guard.py
index 83e98565f9..ce51a08cce 100644
--- a/tests/test_live_system_guard.py
+++ b/tests/test_live_system_guard.py
@@ -37,3 +37,29 @@ def test_wrapped_killer_command_is_still_blocked():
 def test_env_wrapped_killer_command_is_still_blocked():
     with pytest.raises(RuntimeError, match="live-system guard"):
         subprocess.run(["env", "GUARD_TEST=1", "pkill", "-f", "hermes-guard-regression-nomatch"])
+
+
+def test_gateway_start_inside_a_container_exec_is_not_blocked():
+    """``docker exec <ctr> hermes gateway start`` launches the gateway INSIDE the container,
+    where it cannot reach the host's systemd unit or webhook port; tests/docker/ depends on it.
+    The binary is a stub so the argv stays exact without needing a Docker daemon."""
+    import os
+    import stat
+
+    stub_dir = os.path.join(os.environ["HERMES_HOME"], "stub-bin")
+    os.makedirs(stub_dir, exist_ok=True)
+    stub = os.path.join(stub_dir, "docker")
+    with open(stub, "w") as fh:
+        fh.write("#!/bin/sh\nexit 0\n")
+    os.chmod(stub, os.stat(stub).st_mode | stat.S_IXUSR)
+    result = subprocess.run(
+        [stub, "exec", "-u", "hermes", "ctr", "sh", "-c", "hermes -p prof gateway start"],
+        capture_output=True,
+        text=True,
+    )
+    assert result.returncode == 0
+
+
+def test_gateway_start_on_the_host_is_still_blocked():
+    with pytest.raises(RuntimeError, match="REAL.*gateway runtime"):
+        subprocess.run(["python", "-m", "hermes_cli.main", "gateway", "start"])