fix(cli): defer AWS serverless support to the backend

Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>

libs/cli/langgraph_cli/deploy.py

@@ -1548,11 +1548,6 @@ def _resolve_or_create(
     )
     if found is not None:
         return found.id, step
-    if ctx.endpoints.is_aws and ctx.deployment_type == "dev_zero":
-        raise click.UsageError(
-            "Serverless deployments are not supported on AWS yet. "
-            "Use --deployment-type dev instead."
-        )
     try:
         created, step = _create_deployment(
             ctx.client,
@@ -1565,6 +1560,14 @@ def _resolve_or_create(
             secrets=ctx.secrets,
         )
     except HostBackendError as err:
+        if err.status_code == 400 and err.detail == (
+            "Scale to zero deployments are not supported in AWS environments"
+        ):
+            raise HostBackendError(
+                f"{err.message}\nUse --deployment-type dev instead.",
+                status_code=err.status_code,
+                detail=err.detail,
+            ) from None
         if _needs_a_listener(err):
             raise ListenerRequiredError(
                 "The image has to come from a registry you manage, so re-run with "
@@ -2219,8 +2222,8 @@ def _deploy_base_options(
             click.option(
                 "--deployment-type",
                 type=click.Choice(["serverless", "dedicated", "dev", "prod"]),
-                default=None,
-                show_default="serverless; dev on AWS",
+                default="serverless",
+                show_default=True,
                 help=(
                     "Deployment type (used when creating a new deployment): "
                     "serverless (scale to zero) or dedicated (prod); "
@@ -2401,7 +2404,7 @@ def _deploy_cmd(
     host_url: str | None,
     api_key: str | None,
     deployment_id: str | None,
-    deployment_type: str | None,
+    deployment_type: str,
     name: str | None,
     agent_id: str | None,
     environment: str | None,
@@ -2493,9 +2496,6 @@ def _deploy_cmd(
     )
 
     client = _create_host_backend_client(host_url, api_key, env_vars=env_vars)
-    deployment_type = deployment_type or (
-        "dev" if client.endpoints.is_aws else "serverless"
-    )
     try:
         tracked_packages = find_tracked_packages(config, config_json) or None
     except Exception as exc:

libs/cli/langgraph_cli/host_backend.py

@@ -44,10 +44,6 @@ class ControlPlaneEndpoints:
             f".{CLOUD_CONTROL_PLANE_HOST}"
         )
 
-    @property
-    def is_aws(self) -> bool:
-        return urlparse(self.control_plane_url).hostname == "aws.api.host.langchain.com"
-
     @classmethod
     def from_control_plane_url(cls, url: str) -> ControlPlaneEndpoints:
         control_plane_url = url.rstrip("/")

libs/cli/tests/unit_tests/cli/test_deploy_command.py

@@ -345,74 +345,37 @@ def deploy_project(tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> DeployPro
 
 
 @pytest.mark.parametrize("build_mode", ["--no-remote", "--remote"])
-@pytest.mark.parametrize("deployment_args", [(), ("--deployment-type", "dev")])
-def test_aws_deploy_uses_dev(
-    deploy_project: DeployProject, build_mode: str, deployment_args: tuple[str, ...]
+@pytest.mark.parametrize(
+    ("detail", "expects_dev_hint"),
+    [
+        ("Scale to zero deployments are not supported in AWS environments", True),
+        ("No eligible cluster is currently available to host this deployment.", False),
+    ],
+)
+def test_create_failure_preserves_backend_error_and_only_hints_for_aws_rejection(
+    deploy_project: DeployProject, build_mode: str, detail: str, expects_dev_hint: bool
 ) -> None:
-    result = deploy_project.run(
-        build_mode, *deployment_args, host_url="https://aws.api.host.langchain.com"
-    )
+    deploy_project.control_plane.create_error = detail
 
-    assert result.exit_code == 0, result.output
+    result = deploy_project.run(build_mode)
+
+    assert result.exit_code == 1, result.output
+    assert detail in result.output
+    assert ("Use --deployment-type dev instead." in result.output) is expects_dev_hint
     assert deploy_project.control_plane.bodies[CREATE_DEPLOYMENT]["source_config"] == {
-        "deployment_type": "dev"
+        "deployment_type": "dev_zero"
     }
+    assert deploy_project.timeline == [LIST_DEPLOYMENTS, CREATE_DEPLOYMENT]
 
 
 @pytest.mark.parametrize("build_mode", ["--no-remote", "--remote"])
-def test_aws_deploy_rejects_serverless_before_creating_or_building(
+def test_aws_serverless_creation_succeeds_when_backend_accepts_it(
     deploy_project: DeployProject, build_mode: str
 ) -> None:
     result = deploy_project.run(
-        build_mode,
-        "--deployment-type",
-        "serverless",
-        host_url="https://aws.api.host.langchain.com",
+        build_mode, host_url="https://aws.api.host.langchain.com"
     )
 
-    assert result.exit_code == 2, result.output
-    assert "Serverless deployments are not supported on AWS yet" in result.output
-    assert "Use --deployment-type dev instead" in result.output
-    assert deploy_project.timeline == [LIST_DEPLOYMENTS]
-
-
-@pytest.mark.parametrize("endpoint_source", ["environment", "dotenv"])
-def test_aws_default_is_resolved_from_langsmith_endpoint(
-    deploy_project: DeployProject,
-    monkeypatch: pytest.MonkeyPatch,
-    endpoint_source: str,
-) -> None:
-    endpoint = "https://aws.api.smith.langchain.com"
-    if endpoint_source == "environment":
-        monkeypatch.setenv("LANGSMITH_ENDPOINT", endpoint)
-    else:
-        Path(".env").write_text(f"LANGSMITH_ENDPOINT={endpoint}\n")
-    result = CliRunner().invoke(
-        cli,
-        [
-            "deploy",
-            "--api-key",
-            "test-key",
-            "--name",
-            "my-app",
-            "--no-input",
-            "--no-wait",
-            "--remote",
-        ],
-    )
-
-    assert result.exit_code == 0, result.output
-    assert deploy_project.control_plane.bodies[CREATE_DEPLOYMENT]["source_config"] == {
-        "deployment_type": "dev"
-    }
-
-
-def test_explicit_host_takes_precedence_over_aws_langsmith_endpoint(
-    deploy_project: DeployProject, monkeypatch: pytest.MonkeyPatch
-) -> None:
-    monkeypatch.setenv("LANGSMITH_ENDPOINT", "https://aws.api.smith.langchain.com")
-    result = deploy_project.run("--remote", host_url=CLOUD_CONTROL_PLANE_URL)
-
     assert result.exit_code == 0, result.output
     assert deploy_project.control_plane.bodies[CREATE_DEPLOYMENT]["source_config"] == {
         "deployment_type": "dev_zero"

libs/cli/tests/unit_tests/test_host_backend.py

@@ -10,32 +10,6 @@ from langgraph_cli.host_backend import (
 )
 
 
-@pytest.mark.parametrize(
-    ("host_url", "langsmith_endpoint", "expected"),
-    [
-        ("https://aws.api.host.langchain.com/", None, True),
-        ("https://AWS.api.host.langchain.com:443", None, True),
-        (None, "https://aws.api.smith.langchain.com", True),
-        (None, None, False),
-        ("https://eu.api.host.langchain.com", None, False),
-        ("https://apac.api.host.langchain.com", None, False),
-        ("https://aws.example.com/api-host", None, False),
-        ("https://aws.api.host.langchain.com.example.com", None, False),
-        (
-            "https://api.host.langchain.com",
-            "https://aws.api.smith.langchain.com",
-            False,
-        ),
-    ],
-)
-def test_aws_detection_uses_the_resolved_control_plane_host(
-    host_url: str | None, langsmith_endpoint: str | None, expected: bool
-) -> None:
-    endpoints = ControlPlaneEndpoints.resolve(host_url, langsmith_endpoint)
-
-    assert endpoints.is_aws is expected
-
-
 def test_constructor_strips_trailing_slash():
     c = HostBackendClient("https://api.example.com/", "key")
     assert c.base_url == "https://api.example.com"