Slack

Run agents, teams, and workflows in Slack with session routing, streaming, file handling, and human-in-the-loop approvals.

The Slack interface routes messages from Slack users and threads to agents, teams, or workflows in AgentOS.

Coda, a code review agent in Slack

The Slack interface provides:

1. Session routing: Slack users and threads map to AgentOS user and session IDs
2. File handling: Uploads and downloads work out of the box
3. Human in the loop: Pause for approvals before sensitive actions
4. Streaming: Responses stream live with task cards showing progress

Add a database and enable history or memory on the agent or team when replies need context from earlier runs.

Quick start

Agent

agent.py

from agno.agent import Agent
from agno.models.openai import OpenAIResponses
from agno.os.app import AgentOS
from agno.os.interfaces.slack import Slack

agent = Agent(name="Support Bot", model=OpenAIResponses(id="gpt-5.4"))

agent_os = AgentOS(
    agents=[agent],
    interfaces=[Slack(agent=agent)],
)
app = agent_os.get_app()

if __name__ == "__main__":
    agent_os.serve(app="agent:app", reload=True)

Team

team.py

from agno.agent import Agent
from agno.team import Team
from agno.models.openai import OpenAIResponses
from agno.os.app import AgentOS
from agno.os.interfaces.slack import Slack

researcher = Agent(name="Researcher", role="Find information", model=OpenAIResponses(id="gpt-5.4"))
writer = Agent(name="Writer", role="Draft responses", model=OpenAIResponses(id="gpt-5.4"))

support_team = Team(
    name="Support Team",
    mode="coordinate",
    members=[researcher, writer],
    model=OpenAIResponses(id="gpt-5.4"),
)

agent_os = AgentOS(
    teams=[support_team],
    interfaces=[Slack(team=support_team)],
)
app = agent_os.get_app()

if __name__ == "__main__":
    agent_os.serve(app="team:app", reload=True)

The team coordinator routes messages to the right specialist. Each thread is one session belonging to the team.

Workflow

workflow.py

from agno.agent import Agent
from agno.workflow import Step, Workflow
from agno.models.openai import OpenAIResponses
from agno.os.app import AgentOS
from agno.os.interfaces.slack import Slack

researcher = Agent(name="Researcher", model=OpenAIResponses(id="gpt-5.4"))
writer = Agent(name="Writer", model=OpenAIResponses(id="gpt-5.4"))

research_flow = Workflow(
    name="Research",
    steps=[
        Step(name="Research", agent=researcher),
        Step(name="Write", agent=writer),
    ],
)

agent_os = AgentOS(
    workflows=[research_flow],
    interfaces=[Slack(workflow=research_flow)],
)
app = agent_os.get_app()

if __name__ == "__main__":
    agent_os.serve(app="workflow:app", reload=True)

Each message triggers the workflow. Task cards show step progress as the workflow moves through stages.

1. Install dependencies

   uv pip install 'agno[os,slack]' openai
   
2. Create Slack App

   Follow the setup guide to create your app and get credentials.

3. Set credentials

   export OPENAI_API_KEY="..."
   export SLACK_TOKEN="xoxb-..."
   export SLACK_SIGNING_SECRET="..."
   
4. Run

   Run the file from the tab you used:

   Agent

   python agent.py
   

   python team.py
   

   python workflow.py
   

See more examples including streaming, teams, workflows, and multiple bots.

Multiple bots

Run multiple agents on the same server, each with its own Slack App. Useful when you need separate bots for different functions (support vs. sales) or different workspaces.

Agent

multi_bot.py

agent_os = AgentOS(
    agents=[ace_agent, dash_agent],
    interfaces=[
        Slack(
            agent=ace_agent,
            prefix="/ace",
            token=getenv("ACE_SLACK_TOKEN"),
            signing_secret=getenv("ACE_SLACK_SIGNING_SECRET"),
        ),
        Slack(
            agent=dash_agent,
            prefix="/dash",
            token=getenv("DASH_SLACK_TOKEN"),
            signing_secret=getenv("DASH_SLACK_SIGNING_SECRET"),
        ),
    ],
)

Team

multi_team.py

agent_os = AgentOS(
    teams=[support_team, research_team],
    interfaces=[
        Slack(
            team=support_team,
            prefix="/support",
            token=getenv("SUPPORT_SLACK_TOKEN"),
            signing_secret=getenv("SUPPORT_SLACK_SIGNING_SECRET"),
        ),
        Slack(
            team=research_team,
            prefix="/research",
            token=getenv("RESEARCH_SLACK_TOKEN"),
            signing_secret=getenv("RESEARCH_SLACK_SIGNING_SECRET"),
        ),
    ],
)

Workflow

multi_workflow.py

agent_os = AgentOS(
    workflows=[review_flow, publish_flow],
    interfaces=[
        Slack(
            workflow=review_flow,
            prefix="/review",
            token=getenv("REVIEW_SLACK_TOKEN"),
            signing_secret=getenv("REVIEW_SLACK_SIGNING_SECRET"),
        ),
        Slack(
            workflow=publish_flow,
            prefix="/publish",
            token=getenv("PUBLISH_SLACK_TOKEN"),
            signing_secret=getenv("PUBLISH_SLACK_SIGNING_SECRET"),
        ),
    ],
)

Each interface mounts on its own prefix. Set each Slack App’s Request URL accordingly (/ace/events, /dash/events, etc.).

Next steps

Setup Guide

Create a Slack App from scratch

Features

Sessions, files, teams, and more

Human-in-the-Loop

Pause for approval before actions

Reference

All parameters and endpoints