AI Chat Command

You can utilize the Run a Program sub-action to execute scripts with Streamer.bot

Prerequisites

1. Install Python
   Ensure you have the latest version of Python 3 installed on Windows
2. Add Python to $PATH
   Make sure you have added python to your PATH environment variable and you are able to execute python from cmd.
   If you installed Python through the Windows Store, this should be done automatically.
3. Install Ollama
   Ensure you have the latest version of Ollama installed on Windows

Instructions

1. Download llama3.1 model
   For this example, we are going to load up the latest llama3.1 model in ollama:
   Command Prompt

   ollama run llama3.1
   

2. Create our Python script
   In our Streamer.bot folder, navigate to data and then create a scripts directory.
   Streamer.bot\data\scripts\ollama-example.py

   import sys
   import ollama
   
   # Load the prompt from command line arguments
   prompt = " ".join(sys.argv[1:])
   
   # Generate text using llama3.1 model
   response = ollama.generate(model="llama3.1", prompt=prompt)
   
   # Print the response as a single line, truncated at 500 characters
   output = response["response"].replace("\n", " ")
   print(output[0:499])
   

3. Install the ollama python package
   From our scripts directory in Windows Explorer, Right-Click and open the current location in terminal.
   Command Prompt

   pip install ollama
   

   
   This will install the ollama python package to allow our script to interact with ollama.
4. Create a Chat Command
   In Streamer.bot, navigate to the Commands tab
   1. Set the Name to anything you'd like, such as !ollama
   2. Add a command to the Commands list: !ollama
   3. Uncheck the Ignore Internal Messages option
      • This will allow us to use our command from the Streamer.bot chat window!
   
   
   Command Configuration
5. Create our Action
   In Streamer.bot, navigate to the Actions tab
   1. Create a new action, we are going to name ours Ollama Example
   2. Add a Trigger, Core > Commands > Command Triggered
   3. Select the !ollama command we just created
6. Configure Sub-Actions
   1. Add a Run a Program sub-action
      • Core > System > Run a Program
      • Target: python
      • Working Directory: path\to\Streamer.bot\data\scripts
      • Arguments: ollama-example.py %rawInput%
        • Note that we are adding the %rawInput% variable as a python argument.
          This contains the command input from the user, and will allow our python script to read it!
      • Wait: 1
        • This must be set to a value greater than 0!
      
      
      Run a Program Sub-Action
   2. Add a Send Message to Channel sub-action
      • Twitch > Chat > Send Message to Channel
      • Choose whether you want to send the message from your Bot or Broadcaster account
      • Set the Message value to %output0%
        • This variable will contain the first line of output from our python script!
      
      
      Twitch Send Message Sub-Action
7. Done!
   When you are finished, everything should look something like this:
   
   Ollama Action Overview
   You can now test out the !ollama command in chat!
   
   
   Ollama Command Chat Preview

Import Code

Import this code into your Streamer.bot to have the command and action (mostly) set up for you!

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

You will still need to create the python script, and update the working directory of the Run a Program sub-action accordingly.

Tips & Tricks

This is just a basic example, but it can be improved in many ways.

• Twitch chat messages are limited to 500 characters or less.
  If your prompt generates more than 500 characaters worth of output, it will currently be truncated.
  The script could be modified to split output into multiple lines, for Streamer.bot to read in separately. Alternatively, Streamer.bot could be used to split the output into separate messages.

Extended Examples