Slack bot service for seamless ChatGPT integration

• Python >= 3.11.0
• Slack App and Bot tokens
• OpenAI account and valid API key

In order to generate OpenAI API key, you will first need to register and create an API key (if you haven't done already) on the official OpenAI website.

You can find more information how to create and configure Slack app, events, permissions and generate tokens on the official Slack documentation.

These are required permissions for your Slack App:

• app_mentions:read
• channels:history
• channels:read
• chat:write
• files:read
• files:write
• groups:history
• im:history
• mpim:history

These are required events for your Slack App:

• app_mention
• message.im

All configurable environment variables can be found in .env.example file:

Image model dall-e-2 requires image size less than or equal to 1024x1024, dall-e-3 requires greater than or equal to 1024x1024

Variables with * prefix are mandatory for running this service

First you should copy .env.example file into new .env file, and fill Slack tokens and OpenAI API key.

(optional) You should create a new python virtual environment for this project. Run following command from the root of this project using python version 3.8 or greater:

python -m venv ./.venv

Activate virtual environment by running one of the following scripts:

# linux
./.venv/bin/activate

# windows
./.venv/Scripts/activate

Then you can install python requirements by running following command:

pip install -r requirements.txt

Finally, you can start the Slack chatbot service:

python main.py

This service has full docker support with provided by Dockerfile.

The easiest way to pull & run docker image is to use already built public image from official DockerHub repository:

docker pull lukamatosevic/chatgpt-slackbot:latest

docker run --env-file .env lukamatosevic/chatgpt-slackbot

Or, you can build the image yourself with docker command:

docker image build --rm -t chatgpt-slackbot .

Then you can start the chatbot service with any of the following commands:

# provide .env file with configured tokens and api key
docker run --env-file .env chatgpt-slackbot

# also, you can set required environment variables as parameters
docker run --env "SLACK_BOT_TOKEN=xoxb-..." \
           --env "SLACK_APP_TOKEN=xapp-..." \
           --env "OPENAI_API_KEY=sk-..." chatgpt-slackbot

This example shows a conversation with ChatGPT Slack app. You can start conversation directly with ChatGPT by sending it a private message, or you can add it to any channel or group conversation. If you are speaking to the ChatGPT directly, you don't need to mention it by @ChatGPT.

The conversation will keep a specified number of messages in history (default: 3) for some time (default: 15 minutes) which are used as context when ChatGPT is generating new responses in the same thread and channel / private / group chat.

Here is shown the possible use case for asking ChatGPT to answer or do something with the parent message in a thread.

You can also ask ChatGPT to generate images by starting the message with image: prefix. You can also use this in a thread where the parent message will be added to the context when prompting the DALL-E 2 OpenAI API endpoint.

If you want to use the power of ChatGPT as a command-line tool, you should check out my other project which allows interaction with ChatGPT using terminal: https://github.com/lmatosevic/chatgpt-cli

ChatGPT-Slackbot is MIT licensed.