langgraph/reference/prebuilt/

[giscus[bot]]

langgraph/reference/prebuilt/

Build language agents as graphs

https://langchain-ai.github.io/langgraph/reference/prebuilt/

[woodswift]

Could you please provide a better solution to use the pre-defined prompt by create_react_agent() interface? For example, as shown below, the variable prompt is a global variable, and it is used internally by the function modify_messages(). It does not follow the best practice of programming. One potential solution is to move prompt inside the function. However, it might limit the reuse ability of the function. Any advice?

>>> from langchain_core.prompts import ChatPromptTemplate
>>> prompt = ChatPromptTemplate.from_messages([
...     ("system", "You are a helpful bot named Fred."),
...     ("placeholder", "{messages}"),
...     ("user", "Remember, always be polite!"),
... ])
>>> def modify_messages(messages: list):
...     # You can do more complex modifications here
...     return prompt.invoke({"messages": messages})
>>>
>>> graph = create_react_agent(model, tools, messages_modifier=modify_messages)

[SvenDuve]

Just in case somebody else is confused, the following example won't work like this:

>>> from typing import TypedDict
>>> prompt = ChatPromptTemplate.from_messages(
...     [
...         ("system", "Today is {today}"),
...         ("placeholder", "{messages}"),
...     ]
... )
>>>
>>> class CustomState(TypedDict):
...     today: str
...     messages: Annotated[list[BaseMessage], add_messages]
...     is_last_step: str
>>>
>>> graph = create_react_agent(model, tools, state_schema=CustomState, state_modifier=prompt)
>>> inputs = {"messages": [("user", "What's today's date? And what's the weather in SF?")], "today": "July 16, 2004"}
>>> for s in graph.stream(inputs, stream_mode="values"):
...     message = s["messages"][-1]
...     if isinstance(message, tuple):
...         print(message)
...     else:
...         message.pretty_print()

you would have to pass the is_last_step=False:

inputs = {"messages": [("user", "What's today's date? And what's the weather in SF?")], "today": "July 16, 2004", "is_last_step":False}

[sepety]

good day!! please help me solve such a simple problem (but I'm confused in the code and don't see any acceptable examples or instructions) - I want to make an agent with the rag_tool tool that will take the kottext from my pinecone index. what methods should I use, maybe there are examples of such a simple agent? I would be extremely grateful!!

[Heiden133]

You can use Pinecone vector store as a retriever and append it to tools.

https://python.langchain.com/api_reference/pinecone/vectorstores/langchain_pinecone.vectorstores.PineconeVectorStore.html#langchain_pinecone.vectorstores.PineconeVectorStore

[jimmyn88]

Hi,

Is it possibile to get a structured output from a ReAct Agent?

At the moment this is my output:

{
  "action": "search",
  "action_input": "what is the temperature in SF"
}'''

[jimmyn88]

At the moment this is my output:
'''JSON
{
"action": "search",
"action_input": "what is the temperature in SF"
}'''

[amandafanny]

the code in ValidationNode can't run well.

File ~/miniconda3/lib/python3.12/site-packages/langgraph/pregel/__init__.py:1929, in Pregel.invoke(self, input, config, stream_mode, output_keys, interrupt_before, interrupt_after, debug, **kwargs)
   1927 else:
   1928     chunks = []
-> 1929 for chunk in self.stream(
   1930     input,
   1931     config,
   1932     stream_mode=stream_mode,
   1933     output_keys=output_keys,
   1934     interrupt_before=interrupt_before,
   1935     interrupt_after=interrupt_after,
   1936     debug=debug,
   1937     **kwargs,
   1938 ):
   1939     if stream_mode == "values":
   1940         latest = chunk
...
--> 643     raise InvalidUpdateError(msg)

InvalidUpdateError: Expected dict, got ('user', 'Select a number, any number')

[agdev]

When using Groq
The code below does not work in ends up in loop.
==============Code===================

from langchain_core.prompts import ChatPromptTemplate
prompt = ChatPromptTemplate.from_messages([
... ("system", "You are a helpful bot named Fred."),
... ("placeholder", "{messages}"),
... ("user", "Remember, always be polite!"),
... ])
def format_for_model(state: AgentState):
... # You can do more complex modifications here
... return prompt.invoke({"messages": state["messages"]})

graph = create_react_agent(model, tools, state_modifier=format_for_model)
inputs = {"messages": [("user", "What's your name? And what's the weather in SF?")]}
for s in graph.stream(inputs, stream_mode="values"):
... message = s["messages"][-1]
... if isinstance(message, tuple):
... print(message)
... else:
... message.pretty_print()
==============End of code ==============