fix: preserve full LLM config across HITL resume for non-OpenAI providers (#4970)

When a flow with @human_feedback(llm=create_llm()) pauses for HITL and
later resumes:

1. The LLM object was being serialized to just a model string via
   _serialize_llm_for_context() (e.g. 'gemini/gemini-3.1-flash-lite-preview')
2. On resume, resume_async() was creating LLM(model=string) with NO
   credentials, project, location, safety_settings, or client_params
3. OpenAI worked by accident (OPENAI_API_KEY from env), but Gemini with
   service accounts broke

This fix:
- Stashes the live LLM object on the wrapper as _hf_llm attribute
- On resume, looks up the method and retrieves the live LLM if available
- Falls back to the serialized string for backward compatibility
- Preserves _hf_llm through FlowMethod wrapper decorators

Co-authored-by: Joao Moura <joao@crewai.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>

lib/crewai/src/crewai/flow/flow.py

@@ -1315,7 +1315,25 @@ class Flow(Generic[T], metaclass=FlowMeta):
         context = self._pending_feedback_context
         emit = context.emit
         default_outcome = context.default_outcome
-        llm = context.llm
+
+        # Try to get the live LLM from the re-imported decorator instead of the
+        # serialized string. When a flow pauses for HITL and resumes (possibly in
+        # a different process), context.llm only contains a model string like
+        # 'gemini/gemini-3-flash-preview'. This loses credentials, project,
+        # location, safety_settings, and client_params. By looking up the method
+        # on the re-imported flow class, we can retrieve the fully-configured LLM
+        # that was passed to the @human_feedback decorator.
+        llm = context.llm  # fallback to serialized string
+        method = self._methods.get(FlowMethodName(context.method_name))
+        if method is not None:
+            live_llm = getattr(method, "_hf_llm", None)
+            if live_llm is not None:
+                from crewai.llms.base_llm import BaseLLM as BaseLLMClass
+
+                # Only use live LLM if it's a BaseLLM instance (not a string)
+                # String values offer no benefit over the serialized context.llm
+                if isinstance(live_llm, BaseLLMClass):
+                    llm = live_llm
 
         # Determine outcome
         collapsed_outcome: str | None = None

lib/crewai/src/crewai/flow/flow_wrappers.py

@@ -75,6 +75,7 @@ class FlowMethod(Generic[P, R]):
             "__is_router__",
             "__router_paths__",
             "__human_feedback_config__",
+            "_hf_llm",  # Live LLM object for HITL resume
         ]:
             if hasattr(meth, attr):
                 setattr(self, attr, getattr(meth, attr))

lib/crewai/src/crewai/flow/human_feedback.py

@@ -572,6 +572,14 @@ def human_feedback(
             wrapper.__is_router__ = True
             wrapper.__router_paths__ = list(emit)
 
+        # Stash the live LLM object for HITL resume to retrieve.
+        # When a flow pauses for human feedback and later resumes (possibly in a
+        # different process), the serialized context only contains a model string.
+        # By storing the original LLM on the wrapper, resume_async can retrieve
+        # the fully-configured LLM (with credentials, project, safety_settings, etc.)
+        # instead of creating a bare LLM from just the model string.
+        wrapper._hf_llm = llm
+
         return wrapper  # type: ignore[no-any-return]
 
     return decorator

lib/crewai/src/crewai/tools/base_tool.py

@@ -281,6 +281,7 @@ class BaseTool(BaseModel, ABC):
             result_as_answer=self.result_as_answer,
             max_usage_count=self.max_usage_count,
             current_usage_count=self.current_usage_count,
+            cache_function=self.cache_function,
         )
         structured_tool._original_tool = self
         return structured_tool

lib/crewai/src/crewai/tools/structured_tool.py

@@ -58,6 +58,7 @@ class CrewStructuredTool:
         result_as_answer: bool = False,
         max_usage_count: int | None = None,
         current_usage_count: int = 0,
+        cache_function: Callable[..., bool] | None = None,
     ) -> None:
         """Initialize the structured tool.
 
@@ -69,6 +70,7 @@ class CrewStructuredTool:
             result_as_answer: Whether to return the output directly
             max_usage_count: Maximum number of times this tool can be used. None means unlimited usage.
             current_usage_count: Current number of times this tool has been used.
+            cache_function: Function to determine if the tool result should be cached.
         """
         self.name = name
         self.description = description
@@ -78,6 +80,7 @@ class CrewStructuredTool:
         self.result_as_answer = result_as_answer
         self.max_usage_count = max_usage_count
         self.current_usage_count = current_usage_count
+        self.cache_function = cache_function
         self._original_tool: BaseTool | None = None
 
         # Validate the function signature matches the schema
@@ -86,7 +89,7 @@ class CrewStructuredTool:
     @classmethod
     def from_function(
         cls,
-        func: Callable,
+        func: Callable[..., Any],
         name: str | None = None,
         description: str | None = None,
         return_direct: bool = False,
@@ -147,7 +150,7 @@ class CrewStructuredTool:
     @staticmethod
     def _create_schema_from_function(
         name: str,
-        func: Callable,
+        func: Callable[..., Any],
     ) -> type[BaseModel]:
         """Create a Pydantic schema from a function's signature.
 
@@ -182,7 +185,7 @@ class CrewStructuredTool:
 
         # Create model
         schema_name = f"{name.title()}Schema"
-        return create_model(schema_name, **fields)  # type: ignore[call-overload]
+        return create_model(schema_name, **fields)  # type: ignore[call-overload, no-any-return]
 
     def _validate_function_signature(self) -> None:
         """Validate that the function signature matches the args schema."""
@@ -210,7 +213,7 @@ class CrewStructuredTool:
                         f"not found in args_schema"
                     )
 
-    def _parse_args(self, raw_args: str | dict) -> dict:
+    def _parse_args(self, raw_args: str | dict[str, Any]) -> dict[str, Any]:
         """Parse and validate the input arguments against the schema.
 
         Args:
@@ -234,8 +237,8 @@ class CrewStructuredTool:
 
     async def ainvoke(
         self,
-        input: str | dict,
-        config: dict | None = None,
+        input: str | dict[str, Any],
+        config: dict[str, Any] | None = None,
         **kwargs: Any,
     ) -> Any:
         """Asynchronously invoke the tool.
@@ -269,7 +272,7 @@ class CrewStructuredTool:
         except Exception:
             raise
 
-    def _run(self, *args, **kwargs) -> Any:
+    def _run(self, *args: Any, **kwargs: Any) -> Any:
         """Legacy method for compatibility."""
         # Convert args/kwargs to our expected format
         input_dict = dict(zip(self.args_schema.model_fields.keys(), args, strict=False))
@@ -277,7 +280,10 @@ class CrewStructuredTool:
         return self.invoke(input_dict)
 
     def invoke(
-        self, input: str | dict, config: dict | None = None, **kwargs: Any
+        self,
+        input: str | dict[str, Any],
+        config: dict[str, Any] | None = None,
+        **kwargs: Any,
     ) -> Any:
         """Main method for tool execution."""
         parsed_args = self._parse_args(input)
@@ -313,9 +319,10 @@ class CrewStructuredTool:
             self._original_tool.current_usage_count = self.current_usage_count
 
     @property
-    def args(self) -> dict:
+    def args(self) -> dict[str, Any]:
         """Get the tool's input arguments schema."""
-        return self.args_schema.model_json_schema()["properties"]
+        schema: dict[str, Any] = self.args_schema.model_json_schema()["properties"]
+        return schema
 
     def __repr__(self) -> str:
         return f"CrewStructuredTool(name='{sanitize_tool_name(self.name)}', description='{self.description}')"

lib/crewai/tests/test_async_human_feedback.py

@@ -1216,3 +1216,275 @@ class TestAsyncHumanFeedbackEdgeCases:
 
             assert flow.last_human_feedback.outcome == "approved"
             assert flow.last_human_feedback.feedback == ""
+
+
+# =============================================================================
+# Tests for _hf_llm attribute and live LLM resolution on resume
+# =============================================================================
+
+
+class TestLiveLLMPreservationOnResume:
+    """Tests for preserving the full LLM config across HITL resume."""
+
+    def test_hf_llm_attribute_set_on_wrapper_with_basellm(self) -> None:
+        """Test that _hf_llm is set on the wrapper when llm is a BaseLLM instance."""
+        from crewai.llms.base_llm import BaseLLM
+
+        # Create a mock BaseLLM object
+        mock_llm = MagicMock(spec=BaseLLM)
+        mock_llm.model = "gemini/gemini-3-flash"
+
+        class TestFlow(Flow):
+            @start()
+            @human_feedback(
+                message="Review:",
+                emit=["approved", "rejected"],
+                llm=mock_llm,
+            )
+            def review(self):
+                return "content"
+
+        flow = TestFlow()
+        method = flow._methods.get("review")
+        assert method is not None
+        assert hasattr(method, "_hf_llm")
+        assert method._hf_llm is mock_llm
+
+    def test_hf_llm_attribute_set_on_wrapper_with_string(self) -> None:
+        """Test that _hf_llm is set on the wrapper even when llm is a string."""
+
+        class TestFlow(Flow):
+            @start()
+            @human_feedback(
+                message="Review:",
+                emit=["approved", "rejected"],
+                llm="gpt-4o-mini",
+            )
+            def review(self):
+                return "content"
+
+        flow = TestFlow()
+        method = flow._methods.get("review")
+        assert method is not None
+        assert hasattr(method, "_hf_llm")
+        assert method._hf_llm == "gpt-4o-mini"
+
+    @patch("crewai.flow.flow.crewai_event_bus.emit")
+    def test_resume_async_uses_live_basellm_over_serialized_string(
+        self, mock_emit: MagicMock
+    ) -> None:
+        """Test that resume_async uses the live BaseLLM from decorator instead of serialized string.
+
+        This is the main bug fix: when a flow resumes, it should use the fully-configured
+        LLM from the re-imported decorator (with credentials, project, etc.) instead of
+        creating a new LLM from just the model string.
+        """
+        with tempfile.TemporaryDirectory() as tmpdir:
+            db_path = os.path.join(tmpdir, "test_flows.db")
+            persistence = SQLiteFlowPersistence(db_path)
+
+            from crewai.llms.base_llm import BaseLLM
+
+            # Create a mock BaseLLM with full config (simulating Gemini with service account)
+            live_llm = MagicMock(spec=BaseLLM)
+            live_llm.model = "gemini/gemini-3-flash"
+
+            class TestFlow(Flow):
+                result_path: str = ""
+
+                @start()
+                @human_feedback(
+                    message="Approve?",
+                    emit=["approved", "rejected"],
+                    llm=live_llm,  # Full LLM object with credentials
+                )
+                def review(self):
+                    return "content"
+
+                @listen("approved")
+                def handle_approved(self):
+                    self.result_path = "approved"
+                    return "Approved!"
+
+            # Save pending feedback with just a model STRING (simulating serialization)
+            context = PendingFeedbackContext(
+                flow_id="live-llm-test",
+                flow_class="TestFlow",
+                method_name="review",
+                method_output="content",
+                message="Approve?",
+                emit=["approved", "rejected"],
+                llm="gemini/gemini-3-flash",  # Serialized string, NOT the live object
+            )
+            persistence.save_pending_feedback(
+                flow_uuid="live-llm-test",
+                context=context,
+                state_data={"id": "live-llm-test"},
+            )
+
+            # Restore flow - this re-imports the class with the live LLM
+            flow = TestFlow.from_pending("live-llm-test", persistence)
+
+            # Mock _collapse_to_outcome to capture what LLM it receives
+            captured_llm = []
+
+            def capture_llm(feedback, outcomes, llm):
+                captured_llm.append(llm)
+                return "approved"
+
+            with patch.object(flow, "_collapse_to_outcome", side_effect=capture_llm):
+                flow.resume("looks good!")
+
+            # The key assertion: _collapse_to_outcome received the LIVE BaseLLM object,
+            # NOT the serialized string. The live_llm was captured at class definition
+            # time and stored on the method wrapper as _hf_llm.
+            assert len(captured_llm) == 1
+            # Verify it's the same object that was passed to the decorator
+            # (which is stored on the method's _hf_llm attribute)
+            method = flow._methods.get("review")
+            assert method is not None
+            assert captured_llm[0] is method._hf_llm
+            # And verify it's a BaseLLM instance, not a string
+            assert isinstance(captured_llm[0], BaseLLM)
+
+    @patch("crewai.flow.flow.crewai_event_bus.emit")
+    def test_resume_async_falls_back_to_serialized_string_when_no_hf_llm(
+        self, mock_emit: MagicMock
+    ) -> None:
+        """Test that resume_async falls back to context.llm when _hf_llm is not available.
+
+        This ensures backward compatibility with flows that were paused before this fix.
+        """
+        with tempfile.TemporaryDirectory() as tmpdir:
+            db_path = os.path.join(tmpdir, "test_flows.db")
+            persistence = SQLiteFlowPersistence(db_path)
+
+            class TestFlow(Flow):
+                @start()
+                @human_feedback(
+                    message="Approve?",
+                    emit=["approved", "rejected"],
+                    llm="gpt-4o-mini",
+                )
+                def review(self):
+                    return "content"
+
+            # Save pending feedback
+            context = PendingFeedbackContext(
+                flow_id="fallback-test",
+                flow_class="TestFlow",
+                method_name="review",
+                method_output="content",
+                message="Approve?",
+                emit=["approved", "rejected"],
+                llm="gpt-4o-mini",
+            )
+            persistence.save_pending_feedback(
+                flow_uuid="fallback-test",
+                context=context,
+                state_data={"id": "fallback-test"},
+            )
+
+            flow = TestFlow.from_pending("fallback-test", persistence)
+
+            # Remove _hf_llm to simulate old decorator without this attribute
+            method = flow._methods.get("review")
+            if hasattr(method, "_hf_llm"):
+                delattr(method, "_hf_llm")
+
+            # Mock _collapse_to_outcome to capture what LLM it receives
+            captured_llm = []
+
+            def capture_llm(feedback, outcomes, llm):
+                captured_llm.append(llm)
+                return "approved"
+
+            with patch.object(flow, "_collapse_to_outcome", side_effect=capture_llm):
+                flow.resume("looks good!")
+
+            # Should fall back to the serialized string
+            assert len(captured_llm) == 1
+            assert captured_llm[0] == "gpt-4o-mini"
+
+    @patch("crewai.flow.flow.crewai_event_bus.emit")
+    def test_resume_async_uses_string_from_context_when_hf_llm_is_string(
+        self, mock_emit: MagicMock
+    ) -> None:
+        """Test that when _hf_llm is a string (not BaseLLM), we still use context.llm.
+
+        String LLM values offer no benefit over the serialized context.llm,
+        so we don't prefer them.
+        """
+        with tempfile.TemporaryDirectory() as tmpdir:
+            db_path = os.path.join(tmpdir, "test_flows.db")
+            persistence = SQLiteFlowPersistence(db_path)
+
+            class TestFlow(Flow):
+                @start()
+                @human_feedback(
+                    message="Approve?",
+                    emit=["approved", "rejected"],
+                    llm="gpt-4o-mini",  # String LLM
+                )
+                def review(self):
+                    return "content"
+
+            # Save pending feedback
+            context = PendingFeedbackContext(
+                flow_id="string-llm-test",
+                flow_class="TestFlow",
+                method_name="review",
+                method_output="content",
+                message="Approve?",
+                emit=["approved", "rejected"],
+                llm="gpt-4o-mini",
+            )
+            persistence.save_pending_feedback(
+                flow_uuid="string-llm-test",
+                context=context,
+                state_data={"id": "string-llm-test"},
+            )
+
+            flow = TestFlow.from_pending("string-llm-test", persistence)
+
+            # Verify _hf_llm is a string
+            method = flow._methods.get("review")
+            assert method._hf_llm == "gpt-4o-mini"
+
+            # Mock _collapse_to_outcome to capture what LLM it receives
+            captured_llm = []
+
+            def capture_llm(feedback, outcomes, llm):
+                captured_llm.append(llm)
+                return "approved"
+
+            with patch.object(flow, "_collapse_to_outcome", side_effect=capture_llm):
+                flow.resume("looks good!")
+
+            # Should use context.llm since _hf_llm is a string (not BaseLLM)
+            assert len(captured_llm) == 1
+            assert captured_llm[0] == "gpt-4o-mini"
+
+    def test_hf_llm_set_for_async_wrapper(self) -> None:
+        """Test that _hf_llm is set on async wrapper functions."""
+        import asyncio
+        from crewai.llms.base_llm import BaseLLM
+
+        mock_llm = MagicMock(spec=BaseLLM)
+        mock_llm.model = "gemini/gemini-3-flash"
+
+        class TestFlow(Flow):
+            @start()
+            @human_feedback(
+                message="Review:",
+                emit=["approved", "rejected"],
+                llm=mock_llm,
+            )
+            async def async_review(self):
+                return "content"
+
+        flow = TestFlow()
+        method = flow._methods.get("async_review")
+        assert method is not None
+        assert hasattr(method, "_hf_llm")
+        assert method._hf_llm is mock_llm

lib/crewai/tests/tools/test_structured_tool.py

@@ -38,6 +38,44 @@ def test_initialization(basic_function, schema_class):
     assert tool.args_schema == schema_class
 
 
+def test_cache_function_passed_through(basic_function, schema_class):
+    """Test that cache_function is stored on CrewStructuredTool."""
+
+    def no_cache(_args: dict, _result: str) -> bool:
+        return False
+
+    tool = CrewStructuredTool(
+        name="test_tool",
+        description="Test tool description",
+        func=basic_function,
+        args_schema=schema_class,
+        cache_function=no_cache,
+    )
+
+    assert tool.cache_function is no_cache
+
+
+def test_base_tool_passes_cache_function_to_structured_tool():
+    """Test that BaseTool.to_structured_tool propagates cache_function."""
+    from crewai.tools import BaseTool
+
+    def no_cache(_args: dict, _result: str) -> bool:
+        return False
+
+    class MyCacheTool(BaseTool):
+        name: str = "cache_test"
+        description: str = "tool for testing cache passthrough"
+
+        def _run(self, query: str = "") -> str:
+            return "result"
+
+    my_tool = MyCacheTool()
+    my_tool.cache_function = no_cache  # type: ignore[assignment]
+    structured = my_tool.to_structured_tool()
+
+    assert structured.cache_function is no_cache
+
+
 def test_from_function(basic_function):
     """Test creating tool from function"""
     tool = CrewStructuredTool.from_function(