├── .gitignore
├── README.md
├── main.py
└── requirements.txt


/.gitignore:
--------------------------------------------------------------------------------
1 | venv


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
 1 | # LangChain-Streamlit Template
 2 | 
 3 | This repo serves as a template for how to deploy a [LangGraph](https://langchain-ai.github.io/langgraph/) agent on Streamlit.
 4 | 
 5 | This repo contains an `main.py` file which has a template for a chatbot implementation.
 6 | 
 7 | ## Adding your chain
 8 | 
 9 | To add your chain, you need to change the `load_chain` function in `main.py`.
10 | Depending on the type of your chain, you may also need to change the inputs/outputs that occur later on.
11 | 
12 | ## Run locally
13 | 
14 | After installing dependencies with e.g. `$ pip install -r requirements.txt`, you can run this project locally with the following command:
15 | 
16 | ```sh
17 | $ streamlit run main.py
18 | ```
19 | 
20 | ## Deploy on Streamlit
21 | 
22 | This is easily deployable on the Streamlit platform.
23 | Note that when setting up your Streamlit app you should make sure to add `OPENAI_API_KEY` as a secret environment variable.
24 | 
25 | ## Setting up LangSmith
26 | 
27 | To quickly spot issues and improve the performance of your LangGraph projects, sign up for [LangSmith](https://docs.smith.langchain.com/). LangSmith lets you use trace data to debug, test, and monitor your LLM apps built with LangGraph — read more about how to get started [here](https://docs.smith.langchain.com/).
28 | 


--------------------------------------------------------------------------------
/main.py:
--------------------------------------------------------------------------------
 1 | """Python file to serve as the frontend"""
 2 | import uuid
 3 | 
 4 | import streamlit as st
 5 | from streamlit_chat import message
 6 | 
 7 | from langgraph.prebuilt import create_react_agent
 8 | from langchain.chat_models import init_chat_model
 9 | from langgraph.checkpoint.memory import MemorySaver
10 | 
11 | def load_agent():
12 |     """Logic for loading the agent you want to use should go here."""
13 |     llm = init_chat_model("gpt-4o-mini", model_provider="openai", temperature=0)
14 |     # You can declare tools here
15 |     # https://python.langchain.com/docs/how_to/custom_tools/
16 |     tools = []
17 |     prompt = "Always respond like a pirate."
18 |     checkpointer = MemorySaver()
19 |     # https://langchain-ai.github.io/langgraph/reference/prebuilt/#langgraph.prebuilt.chat_agent_executor.create_react_agent
20 |     graph = create_react_agent(
21 |         llm,
22 |         tools=tools,
23 |         state_modifier=prompt,
24 |         checkpointer=checkpointer
25 |     )
26 |     return graph
27 | 
28 | # From here down is all the StreamLit UI.
29 | st.set_page_config(page_title="LangChain Demo", page_icon=":robot:")
30 | st.header("LangChain Demo")
31 | 
32 | if "generated" not in st.session_state:
33 |     st.session_state["generated"] = []
34 | 
35 | if "past" not in st.session_state:
36 |     st.session_state["past"] = []
37 | 
38 | # Add thread_id and checkpointer initialization
39 | if "thread_id" not in st.session_state:
40 |     st.session_state["thread_id"] = str(uuid.uuid4())
41 | 
42 | if "agent" not in st.session_state:
43 |     st.session_state["agent"] = load_agent()
44 | 
45 | def get_text():
46 |     input_text = st.text_input("You: ", "Hello, how are you?", key="input")
47 |     return input_text
48 | 
49 | 
50 | user_input = get_text()
51 | 
52 | if user_input:
53 |     output = st.session_state["agent"].invoke({
54 |         "messages": [{"role": "user", "content": user_input}]
55 |     }, {
56 |         "configurable": {"thread_id": st.session_state.thread_id}
57 |     })
58 | 
59 |     st.session_state.past.append(user_input)
60 |     # Final state in the messages are the output
61 |     st.session_state.generated.append(output["messages"][-1].content)
62 | 
63 | if st.session_state["generated"]:
64 | 
65 |     for i in range(len(st.session_state["generated"]) - 1, -1, -1):
66 |         message(st.session_state["generated"][i], key=str(i))
67 |         message(st.session_state["past"][i], is_user=True, key=str(i) + "_user")
68 | 


--------------------------------------------------------------------------------
/requirements.txt:
--------------------------------------------------------------------------------
1 | langchain
2 | langgraph
3 | langchain_openai
4 | streamlit
5 | streamlit-chat
6 | 


--------------------------------------------------------------------------------