tools.py

from langchain.tools import tool
import contextlib
import io
import traceback

@tool
def execute_python_code(code: str) -> str:
    """
    Executes the given Python code and returns stdout or any error.
    """
    buffer = io.StringIO()
    try:
        with contextlib.redirect_stdout(buffer):
            exec(code, {})
        return buffer.getvalue() or "✅ Code executed successfully with no output."
    except Exception as e:
        return "❌ Execution Error:\n" + traceback.format_exc()
@tool
def web_search(query: str) -> str:
    """Perform a web search using the Tavily API and return the top results's content."""
    from tavily import TavilyClient
    tavily = TavilyClient(api_key=os.environ["tavily_api_key"])
    try:
        return tavily.search(query=query)['results'][0]['content']
    except Exception as e:
        return f"Error from Tavily: {e}"

@tool
def deep_think(prompt: str) -> str:
    """Use an LLM to generate a deeply reasoned response to a prompt."""
    llm = ChatTogether(
        model="meta-llama/Meta-Llama-3-8B-Instruct",
        together_api_key=os.environ["together_api_key"]
    )
    # system_prompt = "You are a thoughtful reasoning assistant. Think deeply and provide insightful reasoning for the input given by the user. And also what are the steps you need to do as per the user input."
    system_prompt = (
    "You are a thoughtful and highly analytical AI assistant trained in critical thinking, planning, and strategy. "
    "Your task is to analyze the user's input carefully and reason through the problem step-by-step. "
    "Start by outlining the problem clearly, identify what is being asked, then break down your reasoning into logical steps. "
    "Also, outline what actions or steps the agent should take based on the prompt — as if you're planning for a human assistant to follow."
)

    response = llm.invoke([
        {"role": "system", "content": system_prompt},
        {"role": "user", "content": prompt}
    ])
    return response.content