Skip to content

Latest commit

 

History

History
42 lines (32 loc) · 1.72 KB

README.md

File metadata and controls

42 lines (32 loc) · 1.72 KB
Raw

Slackbot that uses langchain under the hood

Deploy on Railway

Local development:

  1. Create your Slackbot first, by following this tutorial: https://slack.dev/bolt-python/tutorial/getting-started (We really need to create a bot template, and will do that in the future)

  2. Then, copy .env.example to .env and fill in the values.

pip install -r requirements-dev.txt

#
# To run it locally (no need to use ngrok or anything)
./run_local.sh

# Now you can talk to the bot on Slack
@chatterbot Tell me a joke.

TODO:

  • Deploy the bot to Railway
  • Make the bot respond with markdown
  • Use Redis for permanent memory
  • Use OpenAI streaming to generate text
  • Create a dev and staging bot that runs the bot with a different name in Slack
  • Switch to use a different type of memory, like token buffer: https://langchain.readthedocs.io/en/latest/modules/memory/types/token_buffer.html
  • Get rid of the error that happens when a system message happens and there isn't a "user" field.
  • Welcome people to the channel via something creative
  • On a regular basis (daily?), have Chatterbot say something interesting
  • Notify the user if GPT is being slow...
  • Usernames and Embed profile information about the people chatting with Chatterbot, if it hasn't already seen them before
  • Use Slack-Machine as the framework
  • Make Chatterbot not respond if it is not being spoken to directly
  • Have a "Bot is typing..." message or some other indicator... (Also turns out this is impossible)
  • Make the bot have a green dot if it's online or a grey dot if not (Turns out this is impossible in the new framework??)