Set up for running or developing locally

Whether you're going to develop the app or just run it locally, you need to follow these steps:

ngrok

ngrok allows you to expose a localhost endpoint publicly Create an account on ngrok https://ngrok.com/

Then brew install ngrok

Follow the instructions on ngrok to add the auth token.

In a terminal tab run ngrok http 8000

Copy the Forwarding address, which will look like https://<random-id-here>.ngrok-free.app

Note: this address will change every time you restart ngrok, and you'll need to keep updating your slash command's destination url in slack. You can get one static domain with a free account in the "Domains" tab - if you choose to do that, remember to set your port to 8000 (ngrok's instructions default to 80).

redis

Install redis-stack: brew tap redis-stack/redis-stack then brew install redis-stack

Follow the instructions here to check and update your $PATH with the correct redis-stack-server version

In another terminal tab, start the redis stack server: redis-stack-server

ELi5 Setup

For the ELI5 dropdown action, you'll need to:

Set up the dropdown actions in slack, once you've made your app (see below for more info)
make an OpenAI account.
Then, add an API key
Store the API key in an environment variable in your ~/.zprofile file like so:
```
export OPENAI_API_KEY="<your key here>"
```
Note: the environment variable name here matters and needs to be exact - this is what the open AI library looks for under the hood.

New OpenAI accounts receive $5 of free credit, and each GPT request is fairly inexpensive so this should be sufficient for the hackathon.

If you want to bypass this functionality and run without an OPENAI_API_KEY set, you can pass a flag when running the app (see "run the app" section, below)

run the app

In another terminal tab: Install python dependencies with pip install -r requirements.txt

Then run the python server: The first time you run, you may want to seed the db with common QP terms and definitions. Use: python3 ./main.py --seed=true

To run without seeding, use python3 ./main.py

To allow the app to run without setting an OPENAI_API_KEY, disabling the eli5 command, use: python3 ./main.py --no-ai=true

Call the app's help for more information python3 ./main.py --help

format the app

Install black via pip install black.

To see if and which files would be reformatted, run: python3 -m black --check .

To format all of the files in the directory, run: python3 -m black .

hot reloading

If you want your server to hot reload when you make changes to main.py, run it with jurigged instead.

First, pip install jurigged. Then run the app python3 -m jurigged -v main.py (do this instead of the command in the previous section).

slack app

Create a slack app:

Create New App -> From scratch -> choose a name and select the relevant workspace
Select your new app, go to Features -> Slash Commands to create the command that will trigger this code:
- Name the command
- Assign your ngrok address to the Request URL
In Settings -> Basic Information -> Install your app, choose the workspace to install it to

Save your changes

You should now be able to trigger your command in slack with /name-you-chose

And see the response "Hello, World! Here is a POST response. It worked!"

set up slack message dropdown options

To enable the ELI5 features, you need to do additional setup.

Go to the settings page for your slack apps and select your app:

In Features > Interactivity & Shortcuts, you need to:

Turn on "Interactivity"
Add your ngrok url to "Request URL"
"Create New Shortcut" twice (In both cases, the Name can be whatever you want, but the callback must match the strings below exactly):
- Name: ELI5, Location: Messages, Callback ID: eli5_me
- Name: ELI5, Location: Messages, Callback ID: eli5_me_privately

You should now be able to click on the 3 dot menu for any message, and select your shortcut. If you don't see yours, select "More Message Shortcuts" and search by name.

Vpet95 / wtf-slack