Python Slack SDK
Go to GitHub

Web Client

The Slack Web API allows you to build applications that interact with Slack in more complex ways than the integrations we provide out of the box.

Access Slack’s API methods requires an OAuth token – see the Tokens & Authentication section for more on how Slack uses OAuth tokens as well as best practices.

Each of these API methods is fully documented on our developer site at https://api.slack.com/

The Python document for this module is available at https://slack.dev/python-slack-sdk/api-docs/slack_sdk/

Messaging

Sending a message

One of the primary uses of Slack is posting messages to a channel using the channel ID or as a DM to another person using their user ID. This method will handle either a channel ID or a user ID passed to the channel parameter.

Note that your app’s bot user needs to be in the channel (otherwise, you will get either not_in_channel or channel_not_found error code). If your app has chat:write.public scope, your app can post messages without joining a channel as long as the channel is public. See chat.postMessage for more info.

import logging
logging.basicConfig(level=logging.DEBUG)

import os
from slack_sdk import WebClient
from slack_sdk.errors import SlackApiError

slack_token = os.environ["SLACK_BOT_TOKEN"]
client = WebClient(token=slack_token)

try:
    response = client.chat_postMessage(
        channel="C0XXXXXX",
        text="Hello from your app! :tada:"
    )
except SlackApiError as e:
    # You will get a SlackApiError if "ok" is False
    assert e.response["error"]    # str like 'invalid_auth', 'channel_not_found'

Sending an ephemeral message, which is only visible to an assigned user in a specified channel, is nearly the same as sending a regular message, but with an additional user parameter.

import os
from slack_sdk import WebClient

slack_token = os.environ["SLACK_BOT_TOKEN"]
client = WebClient(token=slack_token)

response = client.chat_postEphemeral(
    channel="C0XXXXXX",
    text="Hello silently from your app! :tada:",
    user="U0XXXXXXX"
)

See chat.postEphemeral for more info.

Formatting with Block Kit

Messages posted from apps can contain more than just text, though. They can include full user interfaces composed of blocks.

The chat.postMessage method takes an optional blocks argument that allows you to customize the layout of a message. Blocks can be specified in a single array of either dict values or slack_sdk.models.blocks.Block objects.

To send a message to a channel, use the channel’s ID. For IMs, use the user’s ID.

client.chat_postMessage(
    channel="C0XXXXXX",
    blocks=[
        {
            "type": "section",
            "text": {
                "type": "mrkdwn",
                "text": "Danny Torrence left the following review for your property:"
            }
        },
        {
            "type": "section",
            "text": {
                "type": "mrkdwn",
                "text": "<https://example.com|Overlook Hotel> \n :star: \n Doors had too many axe holes, guest in room " +
                    "237 was far too rowdy, whole place felt stuck in the 1920s."
            },
            "accessory": {
                "type": "image",
                "image_url": "https://images.pexels.com/photos/750319/pexels-photo-750319.jpeg",
                "alt_text": "Haunted hotel image"
            }
        },
        {
            "type": "section",
            "fields": [
                {
                    "type": "mrkdwn",
                    "text": "*Average Rating*\n1.0"
                }
            ]
        }
    ]
)

Note: You can use the Block Kit Builder to prototype your message’s look and feel.

Threading Messages

Threaded messages are a way of grouping messages together to provide greater context. You can reply to a thread or start a new threaded conversation by simply passing the original message’s ts ID in the thread_ts attribute when posting a message. If you’re replying to a threaded message, you’ll pass the thread_ts ID of the message you’re replying to.

A channel or DM conversation is a nearly linear timeline of messages exchanged between people, bots, and apps. When one of these messages is replied to, it becomes the parent of a thread. By default, threaded replies do not appear directly in the channel, instead relegated to a kind of forked timeline descending from the parent message.

response = client.chat_postMessage(
    channel="C0XXXXXX",
    thread_ts="1476746830.000003",
    text="Hello from your app! :tada:"
)

By default, reply_broadcast is set to False. To indicate your reply is germane to all members of a channel, and therefore a notification of the reply should be posted in-channel, set the reply_broadcast to True.

response = client.chat_postMessage(
    channel="C0XXXXXX",
    thread_ts="1476746830.000003",
    text="Hello from your app! :tada:",
    reply_broadcast=True
)

Note: While threaded messages may contain attachments and message buttons, when your reply is broadcast to the channel, it’ll actually be a reference to your reply, not the reply itself. So, when appearing in the channel, it won’t contain any attachments or message buttons. Also note that updates and deletion of threaded replies works the same as regular messages.

See the Threading messages together article for more information.

Updating a message

Let’s say you have a bot which posts the status of a request. When that request changes, you’ll want to update the message to reflect it’s state.

response = client.chat_update(
    channel="C0XXXXXX",
    ts="1476746830.000003",
    text="updates from your app! :tada:"
)

See chat.update for formatting options and some special considerations when calling this with a bot user.

Deleting a message

Sometimes you need to delete things.

response = client.chat_delete(
    channel="C0XXXXXX",
    ts="1476745373.000002"
)

See chat.delete for more info.

Emoji reactions

You can quickly respond to any message on Slack with an emoji reaction. Reactions can be used for any purpose: voting, checking off to-do items, showing excitement -— or just for fun.

This method adds a reaction (emoji) to an item (file, file comment, channel message, group message, or direct message). One of file, file_comment, or the combination of channel and timestamp must be specified. Also, note that your app’s bot user needs to be in the channel (otherwise, you will get either not_in_channel or channel_not_found error code).

response = client.reactions_add(
    channel="C0XXXXXXX",
    name="thumbsup",
    timestamp="1234567890.123456"
)

Removing an emoji reaction is basically the same format, but you’ll use reactions.remove instead of reactions.add

response = client.reactions_remove(
    channel="C0XXXXXXX",
    name="thumbsup",
    timestamp="1234567890.123456"
)

See reactions.add and reactions.remove for more info.


Files

Uploading files

You can upload files onto Slack and share the file with people in channels. Note that your app’s bot user needs to be in the channel (otherwise, you will get either not_in_channel or channel_not_found error code).

response = client.files_upload(
    channels="C3UKJTQAC",
    file="files.pdf",
    title="Test upload"
)

See files.upload for more info.

Adding a remote file

You can add a file information that is stored in an external storage, not in Slack.

response = client.files_remote_add(
    external_id="the-all-hands-deck-12345",
    external_url="https://{your domain}/files/the-all-hands-deck-12345",
    title="The All-hands Deck",
    preview_image="./preview.png" # will be displayed in channels
)

See files.remote.add for more info.


Conversations

The Slack Conversations API provides your app with a unified interface to work with all the channel-like things encountered in Slack; public channels, private channels, direct messages, group direct messages, and our newest channel type, Shared Channels.

See Conversations API docs for more info.

Start a direct message

The conversations_open method opens either a 1:1 direct message with a single user or a a multi-person direct message, depending on the number of users supplied to the users parameter.

For public or private channels, use the conversations_create method.

Provide a users parameter as an array with 1 to 8 user IDs to open or resume a conversation. Providing only 1 ID will create a direct message. Providing more will create a new multi-party DM or resume an existing conversation.

Subsequent calls to conversations_open with the same set of users will return the already existing conversation.

import os
from slack_sdk import WebClient

client = WebClient(token=os.environ["SLACK_BOT_TOKEN"])
response = client.conversations_open(users=["W123456789", "U987654321"])

See conversations.open additional info.

Creating channels

Creates a new channel, either public or private. The name parameter is required, may contain numbers, letters, hyphens, and underscores, and must contain fewer than 80 characters. To make the channel private, set the option is_private parameter to True.

import os
from slack_sdk import WebClient
from time import time

client = WebClient(token=os.environ["SLACK_BOT_TOKEN"])
channel_name = f"my-private-channel-{round(time())}"
response = client.conversations_create(
    name=channel_name,
    is_private=True
)
channel_id = response["channel"]["id"]
response = client.conversations_archive(channel=channel_id)

See conversations.create additional info.

Listing conversations

To get a list of all the conversations in a workspace, use conversations_list. By default, only public conversations are returned; use the types parameter specify which types of conversations you’re interested in (Note: types is a string of comma-separated values)

import os
from slack_sdk import WebClient

client = WebClient(token=os.environ["SLACK_BOT_TOKEN"])
response = client.conversations_list()
conversations = response["channels"]

Use the types parameter to request additional channels, including public_channel, private_channel, mpim, and im. This parameter is a string of comma-separated values.

import os
from slack_sdk import WebClient

client = WebClient(token=os.environ["SLACK_BOT_TOKEN"])
response = client.conversations_list(
    types="public_channel, private_channel"
)

See conversations.list for more info.

Archived channels are included by default. You can exclude them by passing exclude_archived=True to your request.

response = client.conversations_list(exclude_archived=True)

See conversations.list for more info.

Getting a conversation information

To retrieve a set of metadata about a channel (public, private, DM, or multi-party DM), use conversations_info. The channel parameter is required and must be a valid channel ID. The optional include_locale boolean parameter will return locale data, which may be useful if you wish to return localized responses. The include_num_members boolean parameter will return the number of people in a channel.

import os
from slack_sdk import WebClient

client = WebClient(token=os.environ["SLACK_BOT_TOKEN"])
response = client.conversations_info(
    channel="C031415926",
    include_num_members=1
)

See conversations.info for more info.

Getting members of a conversation

To get a list of the members of a conversation, use conversations_members with the required channel parameter.

import os
from slack_sdk import WebClient

client = WebClient(token=os.environ["SLACK_BOT_TOKEN"])
response = client.conversations_members(channel="C16180339")
user_ids = response["members"]

See conversations.members for more info.

Joining a conversation

Channels are the social hub of most Slack teams. Here’s how you hop into one:

response = client.conversations_join(channel="C0XXXXXXY")

If you are already in the channel, the response is slightly different. already_in_channel will be true, and a limited channel object will be returned. Bot users cannot join a channel on their own, they need to be invited by another user.

See conversations.join for more info.

Leaving a conversation

To leave a conversation, use conversations_leave with the required channel param containing the ID of the channel to leave.

import os
from slack_sdk import WebClient

client = WebClient(token=os.environ["SLACK_BOT_TOKEN"])
response = client.conversations_leave(channel="C27182818")

See conversations.leave for more info.


Modals

Opening a modal

Modals allow you to collect data from users and display dynamic information in a focused surface.

Modals use the same blocks that compose messages with the addition of an input block.

from slack_sdk.signature import SignatureVerifier
signature_verifier = SignatureVerifier(os.environ["SLACK_SIGNING_SECRET"])

from flask import Flask, request, make_response, jsonify
app = Flask(__name__)

@app.route("/slack/events", methods=["POST"])
def slack_app():
    if not signature_verifier.is_valid_request(request.get_data(), request.headers):
        return make_response("invalid request", 403)

    if "payload" in request.form:
        payload = json.loads(request.form["payload"])
        if payload["type"] == "shortcut" and payload["callback_id"] == "test-shortcut":
            # Open a new modal by a global shortcut
            try:
                api_response = client.views_open(
                    trigger_id=payload["trigger_id"],
                    view={
                        "type": "modal",
                        "callback_id": "modal-id",
                        "title": {
                            "type": "plain_text",
                            "text": "Awesome Modal"
                        },
                        "submit": {
                            "type": "plain_text",
                            "text": "Submit"
                        },
                        "blocks": [
                            {
                                "type": "input",
                                "block_id": "b-id",
                                "label": {
                                    "type": "plain_text",
                                    "text": "Input label",
                                },
                                "element": {
                                    "action_id": "a-id",
                                    "type": "plain_text_input",
                                }
                            }
                        ]
                    }
                )
                return make_response("", 200)
            except SlackApiError as e:
                code = e.response["error"]
                return make_response(f"Failed to open a modal due to {code}", 200)

        if (
            payload["type"] == "view_submission"
            and payload["view"]["callback_id"] == "modal-id"
        ):
            # Handle a data submission request from the modal
            submitted_data = payload["view"]["state"]["values"]
            print(submitted_data)    # {'b-id': {'a-id': {'type': 'plain_text_input', 'value': 'your input'}}}

            # Close this modal with an empty response body
            return make_response("", 200)

    return make_response("", 404)

if __name__ == "__main__":
    # export SLACK_SIGNING_SECRET=***
    # export SLACK_BOT_TOKEN=xoxb-***
    # export FLASK_ENV=development
    # python3 app.py
    app.run("localhost", 3000)

See views.open more details and additional parameters.

Also, to run the above example, the following Slack app configurations are required.

  • Enable Interactivity with a valid Request URL: https://{your-public-domain}/slack/events

  • Add a global shortcut with the Callback ID: open-modal-shortcut

Updating and pushing modals

In response to view_submission requests, you can tell Slack to update the current modal view by having “response_action”: “update” and an updated view. Also, there are other response_action types such as errors and push. Refer to the API document for more details.

if (
    payload["type"] == "view_submission"
    and payload["view"]["callback_id"] == "modal-id"
):
    # Handle a data submission request from the modal
    submitted_data = payload["view"]["state"]["values"]
    print(submitted_data)    # {'b-id': {'a-id': {'type': 'plain_text_input', 'value': 'your input'}}}

    # Update the modal with a new view
    return make_response(
        jsonify(
            {
                "response_action": "update",
                "view": {
                    "type": "modal",
                    "title": {"type": "plain_text", "text": "Accepted"},
                    "close": {"type": "plain_text", "text": "Close"},
                    "blocks": [
                        {
                            "type": "section",
                            "text": {
                                "type": "plain_text",
                                "text": "Thanks for submitting the data!",
                            },
                        }
                    ],
                },
            }
        ),
        200,
    )

If your app modify the current modal view when receiving block_actions requests from Slack, you can call views.update API method with the given view ID.

private_metadata = "any str data you want to store"
response = client.views_update(
    view_id=payload["view"]["id"],
    hash=payload["view"]["hash"],
    view={
        "type": "modal",
        "callback_id": "modal-id",
        "private_metadata": private_metadata,
        "title": {
            "type": "plain_text",
            "text": "Awesome Modal"
        },
        "submit": {
            "type": "plain_text",
            "text": "Submit"
        },
        "close": {
            "type": "plain_text",
            "text": "Cancel"
        },
        "blocks": [
            {
                "type": "input",
                "block_id": "b-id",
                "label": {
                    "type": "plain_text",
                    "text": "Input label",
                },
                "element": {
                    "action_id": "a-id",
                    "type": "plain_text_input",
                }
            }
        ]
    }
)

See views.update for more info.

If you want to push a new view onto the modal instead of updating an existing view, reference the views.push documentation.


Rate Limits

When posting messages to a channel, Slack allows applications to send no more than one message per channel per second. We allow bursts over that limit for short periods. However, if your app continues to exceed the limit over a longer period of time it will be rate limited. Different API methods have other rate limits – be sure to check the limits and test that your application has a graceful fallback if it should hit those limits.

If you go over these limits, Slack will start returning a HTTP 429 Too Many Requests error, a JSON object containing the number of calls you have been making, and a Retry-After header containing the number of seconds until you can retry.

Here’s a very basic example of how one might deal with rate limited requests.

import os
import time
from slack_sdk import WebClient
from slack_sdk.errors import SlackApiError

client = WebClient(token=os.environ["SLACK_BOT_TOKEN"])

# Simple wrapper for sending a Slack message
def send_slack_message(channel, message):
    return client.chat_postMessage(
        channel=channel,
        text=message
    )

# Make the API call and save results to `response`
channel = "#random"
message = "Hello, from Python!"
# Do until being rate limited
while True:
    try:
        response = send_slack_message(channel, message)
    except SlackApiError as e:
        if e.response.status_code == 429:
            # The `Retry-After` header will tell you how long to wait before retrying
            delay = int(e.response.headers['Retry-After'])
            print(f"Rate limited. Retrying in {delay} seconds")
            time.sleep(delay)
            response = send_slack_message(channel, message)
        else:
            # other errors
            raise e

Since v3.9.0, the built-in RateLimitErrorRetryHandler is available as an easier way to do the retries for rate limited errors. Refer to the RetryHandler section in this page for more details.

To learn the Slack rate limits in general, see the documentation on Rate Limiting.


Calling any API methods

This library covers all the public endpoints as the methods in WebClient. That said, you may see a bit delay of the library release. When you’re in a hurry, you can directly use api_call method as below.

import os
from slack_sdk import WebClient

client = WebClient(token=os.environ['SLACK_BOT_TOKEN'])
response = client.api_call(
    api_method='chat.postMessage',
    params={'channel': '#random','text': "Hello world!"}
)
assert response["message"]["text"] == "Hello world!"

AsyncWebClient

All the API methods are available in asynchronous programming using the standard asyncio library. You use AsyncWebClient instead for it.

AsyncWebClient internally relies on AIOHTTP library but it is an optional dependency. So, to use this class, run pip install aiohttp beforehand.

import asyncio
import os
# requires: pip install aiohttp
from slack_sdk.web.async_client import AsyncWebClient
from slack_sdk.errors import SlackApiError

client = AsyncWebClient(token=os.environ['SLACK_API_TOKEN'])

# This must be an async method
async def post_message():
    try:
        # Don't forget `await` keyword here
        response = await client.chat_postMessage(
            channel='#random',
            text="Hello world!"
        )
        assert response["message"]["text"] == "Hello world!"
    except SlackApiError as e:
        assert e.response["ok"] is False
        assert e.response["error"]  # str like 'invalid_auth', 'channel_not_found'
        print(f"Got an error: {e.response['error']}")

# This is the simplest way to run the async method
# but you can go with any ways to run it
asyncio.run(post_message())

RetryHandler

With the default settings, only ConnectionErrorRetryHandler with its default configuration (=only one retry in the manner of exponential backoff and jitter) is enabled. The retry handler retries if an API client encounters a connectivity-related failure (e.g., Connection reset by peer).

To use other retry handlers, you can pass a list of RetryHandler to the client constructor. For instance, you can add the built-in RateLimitErrorRetryHandler this way:

import os
from slack_sdk.web import WebClient
client = WebClient(token=os.environ["SLACK_BOT_TOKEN"])

# This handler does retries when HTTP status 429 is returned
from slack_sdk.http_retry.builtin_handlers import RateLimitErrorRetryHandler
rate_limit_handler = RateLimitErrorRetryHandler(max_retry_count=1)

# Enable rate limited error retries as well
client.retry_handlers.append(rate_limit_handler)

Creating your own ones is also quite simple. Defining a new class that inherits slack_sdk.http_retry.RetryHandler (AsyncRetryHandler for asyncio apps) and implements required methods (internals of can_retry / prepare_for_next_retry). Check the built-in ones’ source code for learning how to properly implement.

import socket
from typing import Optional
from slack_sdk.http_retry import (RetryHandler, RetryState, HttpRequest, HttpResponse)
from slack_sdk.http_retry.builtin_interval_calculators import BackoffRetryIntervalCalculator
from slack_sdk.http_retry.jitter import RandomJitter

class MyRetryHandler(RetryHandler):
    def _can_retry(
        self,
        *,
        state: RetryState,
        request: HttpRequest,
        response: Optional[HttpResponse] = None,
        error: Optional[Exception] = None
    ) -> bool:
        # [Errno 104] Connection reset by peer
        return error is not None and isinstance(error, socket.error) and error.errno == 104

client = WebClient(
    token=os.environ["SLACK_BOT_TOKEN"],
    retry_handlers=[MyRetryHandler(
        max_retry_count=1,
        interval_calculator=BackoffRetryIntervalCalculator(
            backoff_factor=0.5,
            jitter=RandomJitter(),
        ),
    )],
)

For asyncio apps, Async prefixed corresponding modules are available. All the methods in those methods are async/await compatible. Check the source code and tests for more details.