This repository contains a C#-based Slack GPT Bot that uses OpenAI's GPT model to answer users' questions. The implementation is based on Slack Sockets API, which means there is no need to host the bot on a server. The bot can be run on any machine.

• Integrate with OpenAI's GPT-4 to answer questions
• Maintain conversation context in a threaded format
• Socket mode integration with Slack
• Splits long messages into multiple messages, and doesn't break the code block formatting
• Parameters for controlling the bot's behavior
• Docker support
• Full documentation

• .NET 7.0
• OpenAI-DotNet
• SlackNet
• SlackNet.AspNetCore

• Restore nuget packages

dotnet restore

1. Go to OpenAI Dashboard
2. Click on "Create new secret key" button.
3. Copy the secret and put it into the OpenAIKey variable in appsettings.json.

NOTE: Using the OpenAI API requires a paid/trial account. You can find more information about pricing here.

Before you can run the Slack GPT Bot, you need to configure the appropriate permissions for your Slack bot. Follow these steps to set up the necessary permissions:

1. Create Slack App
2. Go to your Slack API Dashboard and click on the app you created for this bot.
3. In the left sidebar, click on "OAuth & Permissions".
4. In the "Scopes" section, you will find two types of scopes: "Bot Token Scopes" and "User Token Scopes". Add the following scopes under "Bot Token Scopes":
   • app_mentions:read: Allows the bot to read mention events.
   • chat:write: Allows the bot to send messages.
   • groups:history: Allows the bot to read messages in private channels.
   • channels:history: Allows the bot to read messages in public channels.
   • Please note that other permissions might be required depending on the bot's location.
5. Scroll up to the "OAuth Tokens for Your Workspace" and click "Install App To Workspace" button. This will generate the SlackBotToken.
6. In the left sidebar, click on "Socket Mode" and enable it. You'll be prompted to "Generate an app-level token to enable Socket Mode". Generate a token named SlackAppToken and add the connections:write scope.
7. In the "Features affected" section of "Socket Mode" page, click "Event Subscriptions" and toggle "Enable Events" to " On". Add app_mention event with the app_mentions:read scope in the "Subscribe to bot events" section below the toggle.

1. Run the project

dotnet run --project Slack-GPT-Socket

or see #Docker for more information on how to host it.

2. Invite the bot to your desired Slack channel.
3. Mention the bot in a message and ask a question. The bot will respond with an answer. You can keep mentioning the bot in the same thread to continue the conversation.

You can start your message with the name of the model without parameters eg:

@Bot turbo How are you today?

Will use gpt-3.5-turbo instead of the default gpt-4 model. See GptClient for more aliases.

FAQ: Fine-tuning requests using parameters

Q1: How do I use the parameters?

A1: You can use these parameters at the beginning of your request to fine-tune the model's output. To utilize them, pass the desired parameter name followed by its value. For example: -maxTokens 50. The request should be followed by your output. eg:

@GPT-4 -maxTokens 50 How are you today?

Please note that the parameters should be separated by a space. Should be at the beginning of the request. Right after the bot's name. And should be followed by a space. For parameters that have spaces in them ".

Q2: What does the -maxTokens parameter do?

A2: The -maxTokens parameter limits the number of tokens (words or word segments) in the generated output. You can set this value by passing an integer to the parameter. Example usage: -maxTokens 100. Default is 2048 tokens. GPT-3.5 has a limit of 4000 and GPT-4 has a limit of 8000 tokens.

Q3: How does the -temperature parameter work?

A3: The -temperature parameter controls the randomness of the model's output. Higher values (e.g., 1.0) make the output more random, while lower values (e.g., 0.1) make it more deterministic. You can set this value by passing a float to the parameter. Example usage: -temperature 0.7. Default is 0.7

Q4: What is the -topP parameter?

A4: The -topP parameter (also known as "nucleus sampling") filters the model's token choices based on cumulative probability. You can set this value by passing a float between 0 and 1 to the parameter. Lower values make the output more focused, while higher values allow for more diversity. Example usage: -topP 0.9. Default is 1.

Q5: How do I use the -presencePenalty parameter?

A5: The -presencePenalty parameter penalizes tokens that are already present in the generated text. A higher value discourages repetition, while a lower value allows for more repetition. You can set this value by passing a float to the parameter. Example usage: -presencePenalty 0.5. Default is 0.

Q6: What does the -frequencyPenalty parameter do?

A6: The -frequencyPenalty parameter discourages the use of tokens that appear frequently in the training data. A higher value will make the output more creative, while a lower value will make it more common. You can set this value by passing a float to the parameter. Example usage: -frequencyPenalty 0.3. Default is 0.

Q7: What is the -model parameter?

A7: The -model parameter allows you to specify the name of the model you want to use for generating the output. You can set this value by passing a string to the parameter. Example usage: -model "gpt-3.5-turbo". Default is gpt-4. Available models:

• gpt-4
• gpt-3.5-turbo

Q8: How do I use the -system parameter?

A8: The -system parameter lets you specify the system message that the model should use. The default message is "You are a helpful assistant. Today is {Current Date}", but you can use anything you want. Eg

@GPT-4 -system "You are a Math tutor, your task is to help but not to provide answers so that the student can think for themselves." I don't know how mutch is 37 * 12, please give me an answer.

You can start the docker container with the following command:

docker run -v ./appsettings.json:/app/appsettings.json --restart always ghcr.io/prographers/slack-gpt:latest

You can also use the docker-compose.yml file to start the container, detached. Docker Compose will automatically pull the image from the GitHub Container Registry, and start the container when that happens. It will use watchtower to do that.

docker-compose up -d

Please remember to put the appsettings.json file in the same directory as the command for both cases.

Both images are not exposed on any port, and cannot be accessed from the outside. The only way to access the container is through the Slack API. The container is also running as a non-root user, and has no access to the host system.