app.py

from langchain_tavily import TavilySearch
from fastapi import FastAPI
import os
from langchain.chat_models import init_chat_model
from langgraph.prebuilt import create_react_agent
from langgraph.checkpoint.memory import MemorySaver
from pydantic import BaseModel
from typing import Optional
from langchain_core.messages import HumanMessage

os.environ["GOOGLE_API_KEY"]=os.getenv("GOOGLE_API_KEY")
if not os.environ.get("GOOGLE_API_KEY"):
    raise ValueError("Google API key not found. Please set the GOOGLE_API_KEY environment variable.")

os.environ["TAVILY_API_KEY"]=os.getenv("TAVILY_API_KEY")
if not os.environ.get("TAVILY_API_KEY"):
    raise ValueError("Tavily API key not found. Please set the TAVILY_API_KEY environment variable.")

search=TavilySearch(max_results=2)

tools=[search]

llm=init_chat_model(model="gemini-2.5-flash",model_provider="google_genai")

memory=MemorySaver()
agent_executor = create_react_agent(llm,tools,checkpointer=memory)

class UserRequest(BaseModel):
    message:str
    thread_id:Optional[str]="default_thread" #Use a default thread_id if none is provided

app = FastAPI(
    title="Agent with tavily server",
    description="An API server for a RAG agent built with LangGraph"
)

@app.post("/invoke")
async def invoke_agent(request: UserRequest):
    #Set up the configuration for memory
    config={"configurable":{"thread_id":request.thread_id}}

    input=request.message

    response=agent_executor.invoke({"messages":[HumanMessage(input)]},config=config)
    return {"response":response["messages"][-1]}

#This part is for local testing, can be removed if using a production server
if __name__=="__main__":
    import uvicorn
    uvicorn.run(app,host="0.0.0.0",port=7860)