Confirmation Toolkit

"""
Confirmation Toolkit
=============================

Human-in-the-Loop: Adding User Confirmation to Tool Calls.
"""

from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.models.openai import OpenAIResponses
from agno.tools.websearch import WebSearchTools
from agno.utils import pprint
from rich.console import Console
from rich.prompt import Prompt

console = Console()

# ---------------------------------------------------------------------------
# Create Agent
# ---------------------------------------------------------------------------
agent = Agent(
    model=OpenAIResponses(id="gpt-5-mini"),
    tools=[WebSearchTools(requires_confirmation_tools=["web_search"])],
    markdown=True,
    db=SqliteDb(db_file="tmp/confirmation_required_toolkit.db"),
)

# ---------------------------------------------------------------------------
# Run Agent
# ---------------------------------------------------------------------------
if __name__ == "__main__":
    run_response = agent.run("What is the current stock price of Apple?")
    if run_response.is_paused:  # Or agent.run_response.is_paused
        for requirement in run_response.active_requirements:
            if requirement.needs_confirmation:
                # Ask for confirmation
                console.print(
                    f"Tool name [bold blue]{requirement.tool_execution.tool_name}({requirement.tool_execution.tool_args})[/] requires confirmation."
                )
                message = (
                    Prompt.ask(
                        "Do you want to continue?", choices=["y", "n"], default="y"
                    )
                    .strip()
                    .lower()
                )

                if message == "n":
                    requirement.reject()
                else:
                    requirement.confirm()

        run_response = agent.continue_run(
            run_id=run_response.run_id,
            requirements=run_response.requirements,
        )
        pprint.pprint_run_response(run_response)