fix(OMN-16761): retire the legacy onex run alias — core owns the run verb

docker/runners/runner-image.lock.json

@@ -1,12 +1,12 @@
 {
   "base_image_digest": "sha256:3ba65aa20f86a0fad9df2b2c259c613df006b2e6d0bfcc8a146afb8c525a9751",
   "gh_version": "2.67.0",
-  "identity_digest": "b733067d7533a7fce63acebc1f515516",
+  "identity_digest": "c896b6c2d2ea576b7056c2d57f0e4d0f",
   "image_version": 7,
   "kubectl_version": "1.32.1",
   "python_version": "3.12",
   "runner_version": "2.336.0",
-  "shared_env_digest": "dd7bb511ac7aa391164c1927",
+  "shared_env_digest": "af3a810f5ded3d0f4612a34c",
   "shared_env_install_args": "--frozen --all-extras --all-groups --no-install-project",
   "uv_version": "0.6.14"
 }

pyproject.toml

@@ -168,7 +168,6 @@ delegate = "omnibase_infra.cli.cli_delegate:delegate_command"
 kafka = "omnibase_infra.cli.cli_kafka:kafka"
 node = "omnibase_infra.cli.cli_node:run_node_by_name"
 occ = "omnibase_infra.cli.cli_occ:occ"
-run = "omnibase_infra.cli.cli_node:run_node_by_name"
 skill = "omnibase_infra.cli.cli_skill:run_skill_by_name"
 
 [project.entry-points."onex.backends"]

src/omnibase_infra/cli/model_receipt_runtime_summary.py

@@ -3,7 +3,7 @@
 
 """Typed receipt ``result`` payload for runs without a handler result (OMN-13094).
 
-When ``onex node``/``onex run`` executes in ``--output receipt`` mode, the
+When ``onex node`` executes in ``--output receipt`` mode, the
 receipt's ``result`` field carries the node's full handler result whenever one
 exists. Runs that produce no handler result (event-only orchestrations) or
 that fail carry this summary model instead, so the receipt's ``result`` is

src/omnibase_infra/cli/omnimarket_drift_guard.py

@@ -41,7 +41,7 @@
 short-circuited to a silent no-op, so the exact regression this module
 exists to catch -- ``omnimarket`` silently reverting from a git co-install to
 completely absent -- fell through the guard undetected. The only symptom was
-a generic, unhelpful ``onex skill``/``onex run`` "Unknown node" error with no
+a generic, unhelpful ``onex skill``/``onex node`` "Unknown node" error with no
 pointer back to this module or the repair command (the OMN-13829 ->
 OMN-14060 -> OMN-14531 recurrence: each time, the venv drifted from
 "installed" to "absent", not merely "stale", and the pre-flight guard's
@@ -189,7 +189,7 @@ def check_omnimarket_drift(
             "omnimarket is NOT INSTALLED from git in this interpreter "
             "(absent, or installed from PyPI/a non-VCS source), but a "
             f"canonical clone exists at $OMNI_HOME/omnimarket (HEAD "
-            f"{canonical[:12]}). 'onex skill'/'onex run'/'onex delegate' "
+            f"{canonical[:12]}). 'onex skill'/'onex node'/'onex delegate' "
             "dispatch for market-provided nodes (e.g. node_aislop_sweep) "
             "will fail with 'Unknown node'. Repair with: "
             "scripts/install-node-skill-package.sh --execute (or "

src/omnibase_infra/cli/receipt_mode.py

@@ -1,7 +1,7 @@
 # SPDX-FileCopyrightText: 2025 OmniNode.ai Inc.
 # SPDX-License-Identifier: MIT
 
-"""Receipt-mode execution for ``onex node`` / ``onex run`` (OMN-13094).
+"""Receipt-mode execution for ``onex node`` (OMN-13094).
 
 Layer A of the skill-output-suppression slice
 (``docs/plans/2026-06-12-skill-output-suppression-plan.md``, Phase 2 item 1):

tests/unit/cli/conftest.py

@@ -6,14 +6,13 @@
 Restores the complete process environment after every CLI test because CLI tests
 exercise process-wide runtime configuration in-process (OMN-15572).
 
-Neutralizes the ``onex node``/``onex run`` omnimarket pre-flight drift guard
+Neutralizes the ``onex node`` omnimarket pre-flight drift guard
 (OMN-14560, mirroring OMN-14531's ``onex skill`` fix) so CLI-wiring tests
 across this directory stay hermetic regardless of the ambient developer
 shell's ``$OMNI_HOME`` and whether this test venv happens to have omnimarket
-co-installed. Three files invoke ``run_node_by_name`` (directly or via the
-``onex run`` alias): ``test_cli_node.py``, ``test_cli_node_receipt.py``, and
-``test_onex_run.py`` -- a per-file fixture would have to be duplicated three
-times, so it lives here instead.
+co-installed. Two files invoke ``run_node_by_name``: ``test_cli_node.py`` and
+``test_cli_node_receipt.py`` -- a per-file fixture would have to be duplicated
+twice, so it lives here instead.
 
 The guard's own behavior (fail-open vs. raise) is tested directly in
 ``test_omnimarket_drift_guard.py``, which imports ``check_omnimarket_drift``

tests/unit/cli/test_cli_node.py

@@ -253,7 +253,7 @@ def test_input_file_not_found_surfaces_cli_error(tmp_path: Path) -> None:
 def test_drift_guard_fires_before_unknown_node_lookup(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
-    """OMN-14560: ``onex node``/``onex run`` now call the same drift guard
+    """OMN-14560: ``onex node`` now calls the same drift guard
     ``onex skill`` was fixed to call under OMN-14531.
 
     Before this fix, ``run_node_by_name`` had ZERO drift-guard wiring: a

tests/unit/cli/test_onex_run_alias_retired.py

@@ -0,0 +1,90 @@
+# SPDX-FileCopyrightText: 2025 OmniNode.ai Inc.
+# SPDX-License-Identifier: MIT
+
+"""The legacy ``onex run`` alias must stay retired (OMN-16761).
+
+Background. OMN-8938 renamed this repo's node runner ``onex run`` -> ``onex node``
+and left ``run`` behind as an alias pointing at the *identical* callable. The
+alias was never retired. omnibase_core 0.46.13 (OMN-16677) then legitimately
+reused the freed name for a different command -- the tier-0 local workflow
+harness -- and core's built-in wins the entry-point race:
+
+    onex.cli extension 'run' conflicts with an existing command, skipping
+
+That warning printed on every single CLI invocation in any environment with both
+packages installed, and ``onex run <node>`` silently stopped running nodes.
+
+``run`` now belongs to omnibase_core. This repo owns ``node``. These tests pin
+that boundary from the side this repo actually controls: they assert what
+``omnibase_infra`` *declares*, not how core's harness behaves. Asserting core's
+flag surface from here would couple this repo to core's CLI internals and break
+on every unrelated harness change -- core tests core's command.
+"""
+
+from __future__ import annotations
+
+import tomllib
+from pathlib import Path
+
+import pytest
+
+pytestmark = pytest.mark.unit
+
+_REPO_ROOT = Path(__file__).resolve().parents[3]
+_NODE_RUNNER_TARGET = "omnibase_infra.cli.cli_node:run_node_by_name"
+
+
+def _onex_cli_entry_points() -> dict[str, str]:
+    """Return this repo's declared ``onex.cli`` entry-point table."""
+    data = tomllib.loads((_REPO_ROOT / "pyproject.toml").read_text(encoding="utf-8"))
+    entry_points = data["project"]["entry-points"]["onex.cli"]
+    assert isinstance(entry_points, dict)
+    return entry_points
+
+
+def test_infra_does_not_declare_an_onex_cli_run_entry_point() -> None:
+    """OMN-16761 AC1: no ``run`` entry point, so no collision warning.
+
+    This is the whole defect in one assertion. omnibase_core owns ``onex run``;
+    a ``run`` key here is shadowed by core's built-in and only produces noise.
+    """
+    entry_points = _onex_cli_entry_points()
+    assert "run" not in entry_points, (
+        "omnibase_infra must not declare an 'onex.cli' entry point named 'run' -- "
+        "omnibase_core 0.46.13+ ships a built-in 'onex run' (the tier-0 workflow "
+        "harness) that always wins the entry-point race, so this declaration is "
+        "dead weight that emits "
+        "\"onex.cli extension 'run' conflicts with an existing command, skipping\" "
+        f"on every CLI invocation. Use 'onex node'. Got: {entry_points!r}"
+    )
+
+
+def test_infra_still_declares_the_node_entry_point() -> None:
+    """OMN-16761 AC3: retiring the alias must not disturb ``onex node``.
+
+    ``node`` is the canonical name from OMN-8938 and the only surviving route to
+    this repo's node runner. core ships no built-in ``node``, so nothing shadows it.
+    """
+    entry_points = _onex_cli_entry_points()
+    assert entry_points.get("node") == _NODE_RUNNER_TARGET, (
+        f"'onex node' must resolve to {_NODE_RUNNER_TARGET!r}; got "
+        f"{entry_points.get('node')!r}"
+    )
+
+
+def test_node_runner_is_reachable_under_exactly_one_name() -> None:
+    """No second alias may reintroduce the collision under a different key.
+
+    Guards the *class* of defect, not just the one instance: any future entry
+    point re-pointed at the node runner would recreate an OMN-8938-style alias.
+    """
+    aliases = sorted(
+        name
+        for name, target in _onex_cli_entry_points().items()
+        if target == _NODE_RUNNER_TARGET
+    )
+    assert aliases == ["node"], (
+        "the node runner must be reachable under exactly one 'onex.cli' name "
+        f"('node'); found {aliases!r}. Aliasing one callable under two CLI names "
+        "is what produced the OMN-16761 collision in the first place."
+    )