fix: xml component working properly

[Cristhianzl]

📝 (XMLAgent.py): Add support for chat history data input in XMLAgentComponent to enhance agent functionality and interaction with Language Model.

[aws-amplify-sa-east-1]

This pull request is automatically being deployed by Amplify Hosting (learn more).

Access this pull request here: https://pr-3822.dmtpw4p5recq1.amplifyapp.com

[ogabrielluiz]

Changing the prompt shouldn't be necessary and is probably not a good idea.

[Cristhianzl]

@ogabrielluiz Actually, I will just remove the first "enter" space to improve the UX.

[zcasanova]

I updated it to work for me. Only thing I"m uncertain of is the chat history:

from langchain.agents import create_xml_agent
from langchain_core.prompts import ChatPromptTemplate, PromptTemplate, HumanMessagePromptTemplate

from langflow.base.agents.agent import LCToolsAgentComponent
from langflow.inputs import MultilineInput
from langflow.inputs.inputs import HandleInput


class XMLAgentComponent(LCToolsAgentComponent):
    display_name: str = "XML Agent"
    description: str = "Agent that uses tools formatting instructions as xml to the Language Model."
    icon = "LangChain"
    beta = True
    name = "XMLAgent"

    inputs = LCToolsAgentComponent._base_inputs + [
        HandleInput(name="llm", display_name="Language Model", input_types=["LanguageModel"], required=True),
        MultilineInput(
            name="chat_history",
            display_name="Chat History",
            value=""),
        MultilineInput(
            name="user_prompt",
            display_name="Prompt",
            value="""
You are a helpful assistant. Help the user answer any questions.

You have access to the following tools:

{tools}

In order to use a tool, you can use <tool></tool> and <tool_input></tool_input> tags. You will then get back a response in the form <observation></observation>

For example, if you have a tool called 'search' that could run a google search, in order to search for the weather in SF you would respond:

<tool>search</tool><tool_input>weather in SF</tool_input>

<observation>64 degrees</observation>

When you are done, respond with a final answer between <final_answer></final_answer>. For example:

<final_answer>The weather in SF is 64 degrees</final_answer>

 Begin!

Previous Conversation:
{chat_history}

Question: {input}
{agent_scratchpad}
            """,
        ),
    ]

    def create_agent_runnable(self):
        messages = [
            ("placeholder", "{chat_history}"),
            HumanMessagePromptTemplate(prompt=PromptTemplate(input_variables=["input"], template=self.user_prompt)),
            ("ai", "{agent_scratchpad}"),
        ]
        prompt = ChatPromptTemplate.from_messages(messages)
        return create_xml_agent(self.llm, self.tools, prompt)

[Cristhianzl]

I updated it to work for me. Only thing I"m uncertain of is the chat history:

from langchain.agents import create_xml_agent
from langchain_core.prompts import ChatPromptTemplate, PromptTemplate, HumanMessagePromptTemplate

from langflow.base.agents.agent import LCToolsAgentComponent
from langflow.inputs import MultilineInput
from langflow.inputs.inputs import HandleInput


class XMLAgentComponent(LCToolsAgentComponent):
    display_name: str = "XML Agent"
    description: str = "Agent that uses tools formatting instructions as xml to the Language Model."
    icon = "LangChain"
    beta = True
    name = "XMLAgent"

    inputs = LCToolsAgentComponent._base_inputs + [
        HandleInput(name="llm", display_name="Language Model", input_types=["LanguageModel"], required=True),
        MultilineInput(
            name="chat_history",
            display_name="Chat History",
            value=""),
        MultilineInput(
            name="user_prompt",
            display_name="Prompt",
            value="""
You are a helpful assistant. Help the user answer any questions.

You have access to the following tools:

{tools}

In order to use a tool, you can use <tool></tool> and <tool_input></tool_input> tags. You will then get back a response in the form <observation></observation>

For example, if you have a tool called 'search' that could run a google search, in order to search for the weather in SF you would respond:

<tool>search</tool><tool_input>weather in SF</tool_input>

<observation>64 degrees</observation>

When you are done, respond with a final answer between <final_answer></final_answer>. For example:

<final_answer>The weather in SF is 64 degrees</final_answer>

 Begin!

Previous Conversation:
{chat_history}

Question: {input}
{agent_scratchpad}
            """,
        ),
    ]

    def create_agent_runnable(self):
        messages = [
            ("placeholder", "{chat_history}"),
            HumanMessagePromptTemplate(prompt=PromptTemplate(input_variables=["input"], template=self.user_prompt)),
            ("ai", "{agent_scratchpad}"),
        ]
        prompt = ChatPromptTemplate.from_messages(messages)
        return create_xml_agent(self.llm, self.tools, prompt)

hi @zcasanova
the chat history takes on count olders chat conversation in that session..
works like a memory.. but you have to add the chat history component on the flow

[anovazzi1]

approve for test running
requested by @Cristhianzl