Inserting Prompt Templates in LangGraph - Need Help...

[mraguth]

I have a few of questions here and please help if you can.

Questions:

1. Can I use multiple prompts in a LangGraph prompt as suggested in the example?
2. How can I debug to determine if my prompt is executing correctly?
3. What is the major difference between prompts and tools?

I am referring to this web link for adding multiple tools. My sample example is provided below. Please advise.

https://python.langchain.com/docs/how_to/migrate_agent/

Example code I got from the above reference:

from langgraph.prebuilt import create_react_agent
from langgraph.prebuilt.chat_agent_executor import AgentState

prompt = ChatPromptTemplate.from_messages(
    [
        ("system", "You are a helpful assistant. Respond only in Spanish."),
        ("placeholder", "{messages}"),
    ]
)


def _modify_state_messages(state: AgentState):
    return prompt.invoke({"messages": state["messages"]}).to_messages() + [
        ("user", "Also say 'Pandamonium!' after the answer.")
    ]


langgraph_agent_executor = create_react_agent(
    model, tools, state_modifier=_modify_state_messages
)


messages = langgraph_agent_executor.invoke({"messages": [("human", query)]})
print(
    {
        "input": query,
        "output": messages["messages"][-1].content,
    }
)

My code sample is below:

prompts = [prompt_repl_lg, query_decomposition,decision_prompt]

def _modify_state_messages(state: AgentState):
    modified_messages = state["messages"]
    for prompt in prompts:
        modified_messages = prompt.invoke({"messages": modified_messages}).to_messages()
    modified_messages.append(("user", "Also say 'Pandamonium!' after the answer."))
    return modified_messages

# Create the agent executor
langgraph_agent_executor = create_react_agent(
    model, tools, state_modifier=_modify_state_messages
)

[vbarda]

hi @mraguth

1. Can I use multiple prompts in a LangGraph prompt as suggested in the example?

What are you trying to achieve? in your example you are invoking multiple prompts in a loop, which will result in only the last prompt being applied

2. How can I debug to determine if my prompt is executing correctly?

You can inspect outputs of each prompt invocation separately (outside of create_react_agent) in a Jupyter notebook or use LangSmith

3. What is the major difference between prompts and tools?

Check out the conceptual guide here: https://python.langchain.com/docs/concepts/#tools and https://python.langchain.com/docs/concepts/#prompt-templates

Hope this helps!

[mraguth]

@vbarda My challenge is that I am using multiple prompts in my LangChain script and I want to transition all of them to LangGraph. Should I combine all my prompts into one? Any advice?