agent/gaia_agent.py

import os
from langchain.agents import initialize_agent, AgentType
from langchain.tools import Tool
from langchain.llms import OpenAI
from tools.wikipedia_tool import wiki_search
from tools.audio_transcriber import transcribe_audio
from tools.file_parser import parse_file_and_summarize
from tools.image_chess_solver import solve_chess_image
from tools.youtube_tool import extract_video_id, get_youtube_transcript


def create_langchain_agent():
    tools = [
        Tool(name="Wikipedia Search", func=wiki_search, description="Search Wikipedia for facts."),
        Tool(name="Transcribe Audio", func=transcribe_audio, description="Transcribe MP3 recordings."),
        Tool(name="Image Chess Solver", func=solve_chess_image, description="Solve chess images."),
        Tool(name="File Parser", func=parse_file_and_summarize, description="Summarize files."),
        Tool(name="Youtube Tool Extract", func=extract_video_id, description="Extract videos ids-"),
        Tool(name="Youtube Tool Transscript", func=get_youtube_transcript, description="Transscript youtube videos-"),
    ]

    llm = OpenAI(temperature=0.3, model_name="gpt-4", openai_api_key=os.getenv("Openai")) # Must be set in your env)  # Or use HuggingFaceHub/llama3
    agent = initialize_agent(
        tools=tools,
        llm=llm,
        agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION,
        verbose=True
    )
    return agent