A simple GPT bot which, when asked directly, participates in the conversation using previous messages in the conversation as context. The bot is built using the OpenAI Python library.
IMPORTANT: The bot sends the last few messages in the conversation (for up to 30 minutes in the past by default) to OpenAI to generate a response. Please make sure you are aware of any privacy implications of that.
Once invited to a channel, the bot keeps track of the current conversation. When it encounters a message mentioning it, it responds using GPT to generate the response.
Note that while the bot keeps track of the current conversation even if it is not directly mentioned in the messages, it does NOT send anything to OpenAI until it is specifically menioned in the conversation.
gptbot1.mov
make install
This copies the bot's config as well as the bot script to the
default config directory: ~/botmand-engines
.
Edit ~/botmand-engines/gptbot.yaml
and set the OPENAI_API_KEY
variable.
Check OpenAI documentation on signing up
for the API and retrieving your API
key.
If botmand
is already running, make it reload its engines:
pkill -HUP botmand
At any point in any channel the bot is listening in (by default, #general), send a message to the bot by @mentioning it and ask it any question. The bot uses the last few messages in the conversation along with the question and uses OpenAI to generate a response.