refactor(agent): standardize memory handling and update chat history logic (#8715)

* update chat history

* update to agents

* Update Simple Agent.json

* update to templates

* ruff errors

* Update agent.py

* Update test_agent_component.py

* [autofix.ci] apply automated fixes

* update templates

* test fix

---------

Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
Co-authored-by: Mike Fortman <michael.fortman@datastax.com>

src/backend/base/langflow/base/agents/agent.py

@@ -139,7 +139,10 @@ class LCAgentComponent(Component):
         if hasattr(self, "system_prompt"):
             input_dict["system_prompt"] = self.system_prompt
         if hasattr(self, "chat_history") and self.chat_history:
-            input_dict["chat_history"] = data_to_messages(self.chat_history)
+            if isinstance(self.chat_history, Data):
+                input_dict["chat_history"] = data_to_messages(self.chat_history)
+            if all(isinstance(m, Message) for m in self.chat_history):
+                input_dict["chat_history"] = data_to_messages([m.to_data() for m in self.chat_history])
 
         if hasattr(self, "graph"):
             session_id = self.graph.session_id

src/backend/base/langflow/components/agents/agent.py

@@ -16,7 +16,7 @@ from langflow.components.langchain_utilities.tool_calling import ToolCallingAgen
 from langflow.custom.custom_component.component import _get_component_toolkit
 from langflow.custom.utils import update_component_build_config
 from langflow.field_typing import Tool
-from langflow.io import BoolInput, DropdownInput, MultilineInput, Output
+from langflow.io import BoolInput, DropdownInput, IntInput, MultilineInput, Output
 from langflow.logging import logger
 from langflow.schema.dotdict import dotdict
 from langflow.schema.message import Message
@@ -27,6 +27,9 @@ def set_advanced_true(component_input):
     return component_input
 
 
+MODEL_PROVIDERS_LIST = ["Anthropic", "Google Generative AI", "Groq", "OpenAI"]
+
+
 class AgentComponent(ToolCallingAgentComponent):
     display_name: str = "Agent"
     description: str = "Define the agent's instructions, then enter a task to complete using tools."
@@ -41,11 +44,11 @@ class AgentComponent(ToolCallingAgentComponent):
             name="agent_llm",
             display_name="Model Provider",
             info="The provider of the language model that the agent will use to generate responses.",
-            options=[*sorted(MODEL_PROVIDERS), "Custom"],
+            options=[*MODEL_PROVIDERS_LIST, "Custom"],
             value="OpenAI",
             real_time_refresh=True,
             input_types=[],
-            options_metadata=[MODELS_METADATA[key] for key in sorted(MODEL_PROVIDERS)] + [{"icon": "brain"}],
+            options_metadata=[MODELS_METADATA[key] for key in MODEL_PROVIDERS_LIST] + [{"icon": "brain"}],
         ),
         *MODEL_PROVIDERS_DICT["OpenAI"]["inputs"],
         MultilineInput(
@@ -55,8 +58,17 @@ class AgentComponent(ToolCallingAgentComponent):
             value="You are a helpful assistant that can use tools to answer questions and perform tasks.",
             advanced=False,
         ),
+        IntInput(
+            name="n_messages",
+            display_name="Number of Chat History Messages",
+            value=100,
+            info="Number of chat history messages to retrieve.",
+            advanced=True,
+            show=True,
+        ),
         *LCToolsAgentComponent._base_inputs,
-        *memory_inputs,
+        # removed memory inputs from agent component
+        # *memory_inputs,
         BoolInput(
             name="add_current_date_tool",
             display_name="Current Date",
@@ -78,6 +90,8 @@ class AgentComponent(ToolCallingAgentComponent):
 
             # Get memory data
             self.chat_history = await self.get_memory_data()
+            if isinstance(self.chat_history, Message):
+                self.chat_history = [self.chat_history]
 
             # Add current date tool if enabled
             if self.add_current_date_tool:
@@ -112,13 +126,11 @@ class AgentComponent(ToolCallingAgentComponent):
             raise
 
     async def get_memory_data(self):
-        memory_kwargs = {
-            component_input.name: getattr(self, f"{component_input.name}") for component_input in self.memory_inputs
-        }
-        # filter out empty values
-        memory_kwargs = {k: v for k, v in memory_kwargs.items() if v is not None}
-
-        return await MemoryComponent(**self.get_base_args()).set(**memory_kwargs).retrieve_messages()
+        return (
+            await MemoryComponent(**self.get_base_args())
+            .set(session_id=self.graph.session_id, order="Ascending", n_messages=self.n_messages)
+            .retrieve_messages()
+        )
 
     def get_llm(self):
         if not isinstance(self.agent_llm, str):

src/backend/base/langflow/components/helpers/memory.py

@@ -76,7 +76,7 @@ class MemoryComponent(Component):
             value=100,
             info="Number of messages to retrieve.",
             advanced=True,
-            show=False,
+            show=True,
         ),
         MessageTextInput(
             name="session_id",
@@ -197,28 +197,33 @@ class MemoryComponent(Component):
 
             stored = await self.memory.aget_messages()
             # langchain memories are supposed to return messages in ascending order
+
             if order == "DESC":
                 stored = stored[::-1]
             if n_messages:
-                stored = stored[:n_messages]
+                stored = stored[-n_messages:] if order == "ASC" else stored[:n_messages]
             stored = [Message.from_lc_message(m) for m in stored]
             if sender_type:
                 expected_type = MESSAGE_SENDER_AI if sender_type == MESSAGE_SENDER_AI else MESSAGE_SENDER_USER
                 stored = [m for m in stored if m.type == expected_type]
         else:
+            # For internal memory, we always fetch the last N messages by ordering by DESC
             stored = await aget_messages(
                 sender=sender_type,
                 sender_name=sender_name,
                 session_id=session_id,
-                limit=n_messages,
+                limit=10000,
                 order=order,
             )
-        self.status = stored
+            if n_messages:
+                stored = stored[-n_messages:] if order == "ASC" else stored[:n_messages]
+
+        # self.status = stored
         return cast(Data, stored)
 
     async def retrieve_messages_as_text(self) -> Message:
         stored_text = data_to_text(self.template, await self.retrieve_messages())
-        self.status = stored_text
+        # self.status = stored_text
         return Message(text=stored_text)
 
     async def retrieve_messages_dataframe(self) -> DataFrame:

src/backend/base/langflow/custom/custom_component/component.py

@@ -215,7 +215,7 @@ class Component(CustomComponent):
         """
         return {
             "_user_id": self.user_id,
-            "_session_id": self.session_id,
+            "_session_id": self.graph.session_id,
             "_tracing_service": self._tracing_service,
         }
 

src/backend/tests/unit/components/agents/test_agent_component.py

@@ -14,7 +14,6 @@ from langflow.base.models.openai_constants import (
 from langflow.components.agents.agent import AgentComponent
 from langflow.components.tools.calculator import CalculatorToolComponent
 from langflow.custom import Component
-from langflow.utils.constants import MESSAGE_SENDER_AI, MESSAGE_SENDER_NAME_AI
 
 from tests.base import ComponentTestBaseWithClient, ComponentTestBaseWithoutClient
 from tests.unit.mock_language_model import MockLanguageModel
@@ -50,9 +49,6 @@ class TestAgentComponent(ComponentTestBaseWithoutClient):
             "system_prompt": "You are a helpful assistant.",
             "tools": [],
             "verbose": True,
-            "session_id": str(uuid4()),
-            "sender": MESSAGE_SENDER_AI,
-            "sender_name": MESSAGE_SENDER_NAME_AI,
         }
 
     async def test_build_config_update(self, component_class, default_kwargs):
@@ -129,7 +125,7 @@ class TestAgentComponentWithClient(ComponentTestBaseWithClient):
             input_value=input_value,
             api_key=api_key,
             model_name="gpt-4o",
-            llm_type="OpenAI",
+            agent_llm="OpenAI",
             temperature=temperature,
             _session_id=str(uuid4()),
         )

src/frontend/tests/core/integrations/Financial Agent.spec.ts

@@ -1,78 +0,0 @@
-import { expect, test } from "@playwright/test";
-import * as dotenv from "dotenv";
-import path from "path";
-import { awaitBootstrapTest } from "../../utils/await-bootstrap-test";
-import { withEventDeliveryModes } from "../../utils/withEventDeliveryModes";
-
-withEventDeliveryModes(
-  "Financial Agent",
-  { tag: ["@release", "@starter-projects"] },
-  async ({ page }) => {
-    test.skip(
-      !process?.env?.TAVILY_API_KEY,
-      "TAVILY_API_KEY required to run this test",
-    );
-
-    test.skip(
-      !process?.env?.SAMBANOVA_API_KEY,
-      "SAMBANOVA_API_KEY required to run this test",
-    );
-
-    await page.goto("/");
-    await awaitBootstrapTest(page);
-
-    await page.getByTestId("side_nav_options_all-templates").click();
-    await page.getByRole("heading", { name: "Financial Agent" }).click();
-
-    await page
-      .getByTestId("popover-anchor-input-api_key")
-      .nth(0)
-      .fill(process.env.TAVILY_API_KEY ?? "");
-
-    for (let i = 0; i < 2; i++) {
-      await page.getByTestId("dropdown_str_agent_llm").nth(i).click();
-      await page.waitForTimeout(500);
-      await page.getByRole("option", { name: "SambaNova" }).click();
-    }
-
-    for (let i = 0; i < 3; i++) {
-      await page
-        .getByTestId("value-dropdown-dropdown_str_model_name")
-        .nth(i)
-        .click();
-      await page.waitForTimeout(500);
-
-      await page.getByRole("option").first().click();
-    }
-
-    for (let i = 1; i <= 3; i++) {
-      await page
-        .getByTestId("popover-anchor-input-api_key")
-        .nth(i)
-        .fill(process.env.SAMBANOVA_API_KEY ?? "");
-    }
-
-    await page.getByTestId("playground-btn-flow-io").click();
-
-    await page
-      .getByTestId("input-chat-playground")
-      .last()
-      .fill("Why did Nvidia stock drop in January?");
-
-    await page.getByTestId("button-send").last().click();
-
-    const stopButton = page.getByRole("button", { name: "Stop" });
-    await stopButton.waitFor({ state: "visible", timeout: 30000 });
-
-    if (await stopButton.isVisible()) {
-      await expect(stopButton).toBeHidden({ timeout: 120000 });
-    }
-
-    const output = await page
-      .getByTestId("div-chat-message")
-      .last()
-      .innerText();
-    expect(output.toLowerCase()).toContain("nvidia");
-    expect(output.length).toBeGreaterThan(100);
-  },
-);

src/frontend/tests/core/integrations/Market Research.spec.ts

@@ -41,17 +41,6 @@ withEventDeliveryModes(
       .nth(0)
       .fill(process.env.TAVILY_API_KEY ?? "");
 
-    //* TODO: Remove these 5 steps once the template is updated *//
-    await page.getByTestId("dropdown-output-openaimodel").click();
-
-    await page
-      .getByTestId("dropdown-item-output-openaimodel-language model")
-      .click();
-
-    await page
-      .getByTestId("handle-structuredoutput-shownode-structured output-right")
-      .click();
-
     await page
       .getByTestId("handle-parser-shownode-data or dataframe-left")
       .click();