slackbot

LLM bots for integration with Slack.

Contents

• Models
• Files
• Development
  • (Dev) Installation
  • (Dev) Operation
  • (Dev) Termination
• Deployment
  • Installation
  • Updates
  • Operation
    • Run once
    • Run continuously
  • Termination
• Slack OAuth Scopes

Models

Source: together.ai

Organization	Model Name	Model String for API	Context length	Type
01.AI	01-ai Yi Chat (34B)	zero-one-ai/Yi-34B-Chat	4096	Chat
Meta	LLaMA-2 Chat (13B)	meta-llama/Llama-2-13b-chat-hf	4096	Chat
mistralai	Mistral (7B) Instruct	mistralai/Mistral-7B-Instruct-v0.1	8192	Chat
Stanford	Alpaca (7B)	togethercomputer/alpaca-7b	2048	Chat
Stability AI	Stable Diffusion 2.1	stabilityai/stable-diffusion-2-1	N/A	Image

Files

• This app is run from src/app.py.
• Global variables such as LLM model types and token limits are configured in src/utils/config.py.
• Model-related functions are contained in src/utils/model_funcs.py.
• Text-formatting functions are contained in src/utils/regex_funcs.py.
• Error logging is handled via src/utils/logger.py, with logs stored in logs/

Development

(Dev) Installation

Linux | macOS

$ git clone git@github:jenx-ai/slackbot
$ python3 -m venv venv
$ source venv/bin/activate
(venv)$ python3 -m pip install --upgrade pip
(venv)$ python3 -m pip install -r requirements.txt

Windows

Note that the virtual machine in deployment use Linux. These Windows instructions are provided for development on local machines only.

$ git clone git@github:jenx-ai/slackbot
$ cd slackbot
$ python -m venv venv
$ venv\scripts\activate
(venv)$ python -m pip install --upgrade pip
(venv)$ python -m pip install -r requirements.txt

(Dev) Operation

Run the slackbot from an active virtual environment.

Linux | macOS

$ source venv/bin/activate
(venv)$ cd src
(venv)$ python3 app.py
Bolt app is running!

Windows

$ venv\Scripts\activate
(venv)$ cd src
(venv)$ python app.py
Bolt app is running!

(Dev) Termination

Stop the app with CTRL+C.

Deployment

slackbot is deployed on an AWS EC2 instance named slackbot-vm-micro-04.

The virtual machine (vm) uses Linux and runs Python 3.9 as the system python. uv is installed system-wide as a pip alternative.

The vm uses SSH keys for authorisation with the JENX-AI GitHub repository. The relevant files are id_rsa and id_rsa.pub, located in the vm's ~/.ssh directory. These files will be accessed automatically when attempting to interact with GitHub; the user will need to provide the SSH key password to perform operations such as cloning or fetching from the remote repository.

Push operations have not been authorised - changes to a repository on vm cannot be pushed to overwrite the remote repository.

Directory structure

The current structure for hosting four baseline chat models is as follows:

Parent directory	Model directory	Repo directory
slackbot-repos
	alpaca-7b	slackbot
	Llama-2-13b-chat-hf	slackbot
	Mistral-7B-Instruct-v0.1	slackbot
	Yi-34B-Chat	slackbot

Image models and specialised models will be added when available.

Installation

$ cd slackbot-repos
[ec2-user@ip-172-XX-XX-161~]$ cd {model-dir}
[ec2-user@ip-172-XX-XX-161~model-dir]$ git clone git@github:jenx-ai/slackbot
Cloning into 'slackbot'...
Enter passphrase for key '/home/ec2-user/.ssh/id_rsa':

Enter the password for the SSH key when prompted. If successful, a similar output to that below will display:

remote: Enumerating objects: 94, done.
remote: Counting objects: 100% (31/31), done.
remote: Compressing objects: 100% (28/28), done.
remote: Total 94 (delta 4), reused 11 (delta 3), pack-reused 63
Receiving objects: 100% (94/94), 29.27 KiB | 4.88 MiB/s, done.
Resolving deltas: 100% (33/33), done.
[ec2-user@ip-172-XX-XX-161~model-dir]$ ls
slackbot

Create a virtual environment using uv, then install dependencies from requirements.txt.

[ec2-user@ip-172-XX-XX-161~model-dir]$ cd slackbot
[ec2-user@ip-172-XX-XX-161~slackbot]$ uv venv
[ec2-user@ip-172-XX-XX-161~slackbot]$ source .venv/bin/activate
(venv)[ec2-user@ip-172-XX-XX-161~slackbot]$ uv pip install --upgrade pip
(venv)[ec2-user@ip-172-XX-XX-161~slackbot]$ uv pip install -r requirements.txt

Create a .env file with required variables. The file should be created at the top level of the slackbot repo directory, alongside the src and venv directories.

Updates

Use standard Git commands to fetch the latest state of the remote repository or pull changes. Enter the password for the SSH key when prompted, as per the installation instructions.

$ cd slackbot-repos
[ec2-user@ip-172-XX-XX-161~]$ cd {model-dir}
[ec2-user@ip-172-XX-XX-161~model-dir]$ cd slackbot
[ec2-user@ip-172-XX-XX-161~slackbot]$ git fetch origin main
[ec2-user@ip-172-XX-XX-161~slackbot]$ git pull origin main

Operation

First navigate to the repository and activate the virtual environment, then change to the src directory.

$ cd slackbot-repos
[ec2-user@ip-172-XX-XX-161~]$ cd {model-dir}
[ec2-user@ip-172-XX-XX-161~model-dir]$ cd slackbot
[ec2-user@ip-172-XX-XX-161~slackbot]$ source .venv/bin/activate
(venv)[ec2-user@ip-172-XX-XX-161~slackbot]$ cd src
(venv)[ec2-user@ip-172-XX-XX-161~src]$

Run once

To run the app on a one-off basis, use the same approach as on a local machine - first activate the virtual environment, then navigate to src and run app.py:

(venv)[ec2-user@ip-172-XX-XX-161~src]$ python3 app.py
⚡Bolt app is running!

Run continuously

To run the app continuously so that it will remain active upon exiting the terminal, use the nohup (no hang-up) command:

(venv)[ec2-user@ip-172-XX-XX-161~src]$ nohup python3 app.py 2>&1 &
nohup: ignoring input and appending output to 'nohup.out'

If the operation is successful, the line nohup: ignoring input and appending output to 'nohup.out' will be output, followed by a line that includes a number inside square brackets, like [2814]. This is the process ID (PID) for that process. Make a note of the PID if you might need to manage this process individually later - but you can access all processes running app.py commands together later as per the instructions in the Termination section.

With the programme running under the nohup command, the terminal can be closed and the programme will continue to run.

Termination

If the app is run on a one-off basis using python3 app.py, stop the app with CTRL+C.

If the app was run using the nohup command, connect to a new terminal window for the vm.

Navigate to the location of the repository - be sure to go to the correct model version.

$ cd slackbot-repos
[ec2-user@ip-172-XX-XX-161~]$ cd {model-dir}
[ec2-user@ip-172-XX-XX-161~model-dir]$ cd slackbot

Use the following command to view a list of live processes:

[ec2-user@ip-172-XX-XX-161~slackbot]$ ps aux | grep app.py

The process ID (PID) numbers for each service will be listed as the first number on each line.

Note the PID for the relevant process for the model to terminate, then run:

[ec2-user@ip-172-XX-XX-161~slackbot]$ kill <PID>

Slack OAuth Scopes

The following bot scopes are required for this implementation:

OAuth Scope	Scope Type	Description
app_mentions:read	Bot	View messages that directly mention @app in conversations that the app is in
chat:write	Bot	Send messages as @app
chat:write:public	Bot	Send messages to channels @app isn't a member of
im:history	Bot	View messages and other content in direct messages that app has been added to
connections:write	App	Route your app’s interactions and event payloads over WebSockets