Collect Structured User Input in Slack

Draft a support ticket from the conversation, then pause with a Slack form so the requester supplies the priority and owning component.

Open engineering tickets from Slack conversations. The agent extracts title and description, then Slack collects priority and component before the tool runs.

hitl_user_input.py
"""
Collect Structured User Input in Slack
======================================

Draft a support ticket from the conversation, then pause with a Slack form so
the requester supplies the priority and owning component.

Prerequisites: SLACK_TOKEN, SLACK_SIGNING_SECRET, OPENAI_API_KEY
Run: .venvs/demo/bin/python cookbook/05_agent_os/17_slack/hitl_user_input.py
Try in Slack: Ask "Open a ticket: checkout returns 500 for an empty cart."
Slack scopes: app_mentions:read, assistant:write, chat:write, im:history
"""

from typing import Literal
from uuid import uuid4

from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.models.openai import OpenAIResponses
from agno.os import AgentOS
from agno.os.interfaces.slack import Slack
from agno.tools import tool

tickets = [
    {
        "id": "SUP-A1B2C3",
        "title": "Checkout 500 when cart is empty",
        "status": "open",
    }
]

@tool
def search_existing_tickets(query: str) -> list[dict[str, str]]:
    """Return open tickets whose title contains the query."""
    normalized = query.lower()
    return [
        ticket
        for ticket in tickets
        if normalized in ticket["title"].lower() and ticket["status"] == "open"
    ]

@tool(requires_user_input=True, user_input_fields=["priority", "component"])
def create_support_ticket(
    title: str,
    description: str,
    priority: Literal["P0", "P1", "P2", "P3"],
    component: str,
) -> str:
    """Create a support ticket after Slack collects its routing fields."""
    ticket_id = f"SUP-{uuid4().hex[:6].upper()}"
    tickets.append({"id": ticket_id, "title": title, "status": "open"})
    return (
        f"Opened {ticket_id}: {title} "
        f"(priority={priority}, component={component}). Description: {description}"
    )

# ---------------------------------------------------------------------------
# Create User-input Slack AgentOS
# ---------------------------------------------------------------------------

db = SqliteDb(
    id="slack-hitl-user-input-db",
    db_file="tmp/slack_hitl_user_input.db",
)

support_agent = Agent(
    id="slack-support-intake-agent",
    name="Slack Support Intake",
    model=OpenAIResponses(id="gpt-5.5"),
    db=db,
    tools=[search_existing_tickets, create_support_ticket],
    instructions=[
        "Search for a duplicate before filing a ticket.",
        "If no duplicate exists, draft a concise title and description.",
        "Call create_support_ticket with the title and description.",
        "Priority and component are excluded from the model-visible schema; "
        "Slack collects them through the tool requirement form.",
    ],
    markdown=True,
)

agent_os = AgentOS(
    id="slack-hitl-user-input-os",
    description="AgentOS collecting required tool fields through a Slack form.",
    db=db,
    agents=[support_agent],
    interfaces=[Slack(agent=support_agent)],
)
app = agent_os.get_app()

# ---------------------------------------------------------------------------
# Run User-input Slack AgentOS
# ---------------------------------------------------------------------------

if __name__ == "__main__":
    agent_os.serve(app=app)

Tickets are stored in a process-local list and reset on restart; SQLite persists AgentOS runs. The source's checkout empty-cart prompt matches a seeded ticket and can exercise duplicate detection instead of opening a form. To try the input form, request a new issue such as “Create a ticket: exported invoice PDFs omit the tax registration number.” The model must choose the ticket tool for the form to appear.

Run the Example

Set up your virtual environment

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

Prepare the Slack app

Follow Slack setup to create and install the app, obtain its bot token and signing secret, and configure its current agent experience. Add the bot scopes listed in this example's source docstring and reinstall after changing scopes. Subscribe to app_mention and message.im; configure interactivity for buttons and forms.

Install ngrok and run ngrok http 7777 in another terminal. Keep the tunnel running. After starting this example's server, use the callback paths listed on this page under your public HTTPS URL and complete Slack's verification challenge. Configure each app separately for a multi-app example.

Streaming requires the corresponding Slack app capability. The current Agno adapter initializes suggested prompts on the legacy assistant_thread_started event; the setup guide explains the new-app limitation. Keep only one standalone example on port 7777 at a time.

Install dependencies

uv pip install -U "agno[os,slack]" openai

Export environment variables

export OPENAI_API_KEY="your_openai_api_key_here"
export SLACK_SIGNING_SECRET="your_slack_signing_secret_here"
export SLACK_TOKEN="your_slack_token_here"

Run the example

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

python hitl_user_input.py

Connect Slack to the running server

Keep Python and ngrok running. In each app's Slack settings, prepend your public HTTPS origin to these paths:

AppEvent subscriptionsInteractivity
Slack app/slack/events/slack/interactions

Complete URL verification, then send a DM or invite the app to a channel and @mention it. Ordinary channel replies require an @mention with the default configuration.

Full source: cookbook/05_agent_os/17_slack/hitl_user_input.py