mindflayer

The Mind Flayer, also known as the Shadow Monster, is a malevolent entity that rules the parallel dimension known as the Upside Down.

This code trains markov models on your organizations public slack messages and provides a simple CLI for posting to a channel.

Installation

git clone git@github.com:WillRaphaelson/mindflayer.git
cd mindflayer/
pip install -r requirements.txt

Configuration

Slack-side

Create an app in your workspace at https://api.slack.com/apps, and add in the following app and user scopes.

Bot Token Scopes

Scope	Description
channels:join	Join public channels in the workspace
channels:manage	Manage public channels that Mindflayer has been added to and create new ones
channels:read	View basic information about public channels in the workspace
chat:write	Send messages as @mindflayer
chat:write.public	Send messages to channels @mindflayer isn't a member of
groups:history	View messages and other content in private channels that Mindflayer has been added to
groups:write	Manage private channels that Mindflayer has been added to and create new ones
im:write	Start direct messages with people
mpim:write	Start group direct messages with people
users:read	View people in the workspace

User Token Scopes

Scope	Description
channels:history	View messages and other content in a user’s public channels

Install the application, and securely store the Bot And User OAuth Access Tokens for use in the config file detailed below.

Local

config.py

A config.py file in the top level directory will provide key configuration variables. The slack bot and app token are take from the app's permissions page in Slack.

SLACK_BOT_TOKEN = "xxxx-xxxxxxxxxxxx-xxxxxxxxxxxx-xxxxxxxxxxxxxxxxxxxxxxxx"
SLACK_USER_TOKEN = "xxxx-xxxxxxxxxxxx-xxxxxxxxxxxx-xxxxxxxxxxxxxxxxxxxxxxxx"
TEST_ENV = "GS6CFBN5N"
PROD_ENV = "CBHJ17SVC"

subdirectories

Subdirectory structure and contents are important:

  .
  ├── ...
  ├── channels               # csv files of channel histories
  │   ├── data.csv           
  │   └── ...
  ├── users                  # txt files of user posts
  │   ├── Will_Arr_U001.txt         
  │   └── ...
  ├── models                 # json files of pickled markov models
  │   ├── U001.json         
  │   └── ...
  └── ...

Usage

The train command will scrape, parse, and train new models based on data from the last num days.

python mindflayer.py train --num 1

The post command will surface randomly generated sentences, and prompt for posting in either the test or prod channels.

python mindflayer.py post --env test
Review Posts:

Will Arr
Thank you for the Software Engineering Intern interviews
accept and post? (y/n):

To generate and post sentences for a specific user, provide the optional --user argument followed by a users slack ID

python mindflayer.py post --env test --user UBGGK785R

To generate and post sentences from the mindflayer itself, user the post-arbitrary argument followed by the --env and --mssg arguments

WARNING: only post messages from the mindflayer when it is hella funny

python mindflayer.py post-arbitrary
--env test
--mssg "okay b4 the holiday party do any1 wanna admit they got a crush on me"

WillRaphaelson / mindflayer