fix(actions): rename generate_next_step to generate_next_steps for task-specific LLM support (#1603)

Fixes a naming mismatch bug where the action `generate_next_step`
(singular) didn't match the Task enum value `generate_next_steps`
(plural). This mismatch prevented task-specific LLM configuration from
working correctly.

When users configured a task-specific LLM with `type:
generate_next_steps` in their config, the runtime would look for
`generate_next_step_llm` (based on the action name) but the config
registered `generate_next_steps_llm`, causing the task to fall back to
the main LLM instead.

Author: Pouyanpi

nemoguardrails/actions/llm/generation.py

@@ -599,7 +599,7 @@ async def _search_flows_index(self, text, max_results):
         return final_results[0:max_results]
 
     @action(is_system_action=True)
-    async def generate_next_step(self, events: List[dict], llm: Optional[BaseLLM] = None):
+    async def generate_next_steps(self, events: List[dict], llm: Optional[BaseLLM] = None):
         """Generate the next step in the current conversation flow.
 
         Currently, only generates a next step after a user intent.

nemoguardrails/llm/filters.py

@@ -81,7 +81,7 @@ def co_v2(
         "GenerateFlowFromInstructionsAction",
         "GenerateFlowFromNameAction",
         "generate_intent_steps_message",
-        "generate_next_step",
+        "generate_next_steps",
         "GenerateUserIntentAction",
         "GenerateValueAction",
         "GetLastUserMessageAction",

nemoguardrails/rails/llm/llm_flows.co

@@ -78,7 +78,7 @@ define flow generate next step
   priority 0.9
 
   user ...
-  execute generate_next_step
+  execute generate_next_steps
 
 
 define parallel extension flow generate bot message

tests/test_configs/with_actions_override/config.py

@@ -25,7 +25,7 @@ async def generate_user_intent():
 
 
 @action(is_system_action=True)
-async def generate_next_step():
+async def generate_next_steps():
     return ActionResult(events=[{"type": "BotIntent", "intent": "respond to question"}])
 
 
@@ -38,5 +38,5 @@ async def generate_bot_message():
 
 def init(app: LLMRails):
     app.register_action(generate_user_intent)
-    app.register_action(generate_next_step)
+    app.register_action(generate_next_steps)
     app.register_action(generate_bot_message)

tests/test_general_instructions.py

@@ -158,7 +158,7 @@ async def test_search_flows_index_is_none():
 
 
 @pytest.mark.asyncio
-async def test_generate_next_step_empty_event_list():
+async def test_generate_next_steps_empty_event_list():
     """Check if we try and search the flows index when None we get None back"""
 
     config = RailsConfig(
@@ -175,7 +175,7 @@ async def test_generate_next_step_empty_event_list():
     )
 
     with pytest.raises(RuntimeError, match="No last user intent found from which to generate next step"):
-        _ = await actions.generate_next_step(events=[])
+        _ = await actions.generate_next_steps(events=[])
 
 
 #

tests/test_task_specific_model.py

@@ -0,0 +1,92 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+
+import pytest
+
+from nemoguardrails import LLMRails, RailsConfig
+from nemoguardrails.imports import check_optional_dependency
+from nemoguardrails.rails.llm.options import GenerationResponse
+
+has_langchain_openai = check_optional_dependency("langchain_openai")
+
+has_openai_key = bool(os.getenv("OPENAI_API_KEY"))
+
+skip_if_no_openai = pytest.mark.skipif(
+    not (has_langchain_openai and has_openai_key),
+    reason="Requires langchain_openai and OPENAI_API_KEY environment variable",
+)
+
+
+@skip_if_no_openai
+def test_task_specific_model_for_generate_user_intent_and_generate_next_steps():
+    config = RailsConfig.from_content(
+        colang_content="""
+            define user express greeting
+              "hi"
+
+            define flow
+              user express greeting
+              bot express greeting
+
+            define bot express greeting
+              "Hello! How can I assist you today?"
+        """,
+        yaml_content="""
+            models:
+              - type: main
+                engine: openai
+                model: gpt-3.5-turbo-instruct
+
+              - type: generate_user_intent
+                engine: openai
+                model: gpt-4o-mini
+
+              - type: generate_next_steps
+                engine: openai
+                model: gpt-4o-mini
+        """,
+    )
+
+    rails = LLMRails(config)
+
+    res = rails.generate(
+        messages=[{"role": "user", "content": "what can you do?"}],
+        options={"log": {"llm_calls": True}},
+    )
+
+    assert isinstance(res, GenerationResponse)
+    assert res.log is not None
+    assert res.log.llm_calls is not None
+    assert len(res.log.llm_calls) > 0
+
+    task_specific_tasks = ["generate_user_intent", "generate_next_steps"]
+
+    generate_user_intent_calls = [call for call in res.log.llm_calls if call.task == "generate_user_intent"]
+    assert len(generate_user_intent_calls) > 0
+    for call in generate_user_intent_calls:
+        assert call.llm_model_name == "gpt-4o-mini"
+        assert call.llm_provider_name == "openai"
+
+    generate_next_steps_calls = [call for call in res.log.llm_calls if call.task == "generate_next_steps"]
+    assert len(generate_next_steps_calls) > 0
+    for call in generate_next_steps_calls:
+        assert call.llm_model_name == "gpt-4o-mini"
+        assert call.llm_provider_name == "openai"
+
+    other_calls = [call for call in res.log.llm_calls if call.task not in task_specific_tasks]
+    for call in other_calls:
+        assert call.llm_model_name == "gpt-3.5-turbo-instruct"

tests/tracing/spans/test_span_v2_otel_semantics.py

@@ -121,7 +121,7 @@ def test_llm_span_has_complete_attributes(self):
 
         rail = ActivatedRail(
             type="dialog",
-            name="generate_next_step",
+            name="generate_next_steps",
             started_at=1.0,
             finished_at=2.0,
             duration=1.0,