Team Tool Confirmation Stream

Team HITL Streaming: Tool on the team itself requiring confirmation.

team_tool_confirmation_stream.py
"""Team HITL Streaming: Tool on the team itself requiring confirmation.

This example demonstrates HITL for tools provided directly to the Team
(not to member agents) in streaming mode. When the team leader decides
to use a tool that requires confirmation, the entire team run pauses
until the human confirms.

Note: For team-level tools (not member agent tools), you can use either
isinstance(event, TeamRunPausedEvent) or event.is_paused since there's
no member agent pause to confuse it with.
"""

from agno.agent import Agent
from agno.db.postgres import PostgresDb
from agno.models.openai import OpenAIResponses
from agno.run.team import RunPausedEvent as TeamRunPausedEvent
from agno.team.team import Team
from agno.tools import tool
from agno.utils import pprint

# ---------------------------------------------------------------------------
# Setup
# ---------------------------------------------------------------------------
db = PostgresDb(db_url="postgresql+psycopg://ai:ai@localhost:5532/ai")


# ---------------------------------------------------------------------------
# Tools
# ---------------------------------------------------------------------------
@tool(requires_confirmation=True)
def approve_deployment(environment: str, service: str) -> str:
    """Approve and execute a deployment to an environment.

    Args:
        environment (str): Target environment (staging, production)
        service (str): Service to deploy
    """
    return f"Deployment of {service} to {environment} approved and executed"


# ---------------------------------------------------------------------------
# Create Members
# ---------------------------------------------------------------------------
research_agent = Agent(
    name="Research Agent",
    role="Researches deployment readiness",
    model=OpenAIResponses(id="gpt-5.2"),
    db=db,
)


# ---------------------------------------------------------------------------
# Create Team
# ---------------------------------------------------------------------------
team = Team(
    name="Release Team",
    members=[research_agent],
    model=OpenAIResponses(id="gpt-5.2"),
    tools=[approve_deployment],
    instructions="You manage releases. Use the approve_deployment tool to deploy services. Call it immediately when asked to deploy.",
    db=db,
)


# ---------------------------------------------------------------------------
# Run Team
# ---------------------------------------------------------------------------
if __name__ == "__main__":
    for run_event in team.run(
        "Check if the auth service is ready and deploy it to staging", stream=True
    ):
        # Use isinstance to check for team's pause event
        if isinstance(run_event, TeamRunPausedEvent):
            print("Team paused - requires confirmation for team-level tool")
            for req in run_event.active_requirements:
                if req.needs_confirmation:
                    print(f"  Tool: {req.tool_execution.tool_name}")
                    print(f"  Args: {req.tool_execution.tool_args}")
                    req.confirm()

            response = team.continue_run(
                run_id=run_event.run_id,
                session_id=run_event.session_id,
                requirements=run_event.requirements,
                stream=True,
            )
            pprint.pprint_run_response(response)

Example behavior

The deployment tool is a demonstration stub that returns a success string. The caller automatically calls req.confirm() for every confirmation requirement; replace that line with an actual approval decision before connecting a deployment service. A new tool call can pause the resumed run again; see Multi-Round User Input for a loop that handles repeated pauses.

Run the Example

Set up your virtual environment

uv venv --python 3.12
source .venv/bin/activate

Install dependencies

uv pip install -U agno "psycopg[binary]" openai sqlalchemy

Export your OpenAI API key

export OPENAI_API_KEY="your_openai_api_key_here"

Run PgVector

docker run -d \
  -e POSTGRES_DB=ai \
  -e POSTGRES_USER=ai \
  -e POSTGRES_PASSWORD=ai \
  -e PGDATA=/var/lib/postgresql \
  -v pgvolume:/var/lib/postgresql \
  -p 5532:5432 \
  --name pgvector \
  agnohq/pgvector:18

Run the example

Save the code above as team_tool_confirmation_stream.py, then run:

python team_tool_confirmation_stream.py

Full source: cookbook/03_teams/20_human_in_the_loop/team_tool_confirmation_stream.py