Inspiration - https://github.com/Echoshard/Gemini_Discordbot
This bot acts as a Chatbot. It uses Google's Gemini API which is free for upto 60 calls/min as of Jan 2023.
The bot can be talked to by mentioning it or DMing it.
Each channel/thread has it's own context and can be erased by using /forget.
Additionally, a persona can be specified while making it forget history.
Due to API restrictions, the bot cannot remember image interactions.
Get a Google Gemini api key from ai.google.dev
Make a discord bot at discord.com/developers/applications
Clone repository
Install all dependencies as specified in requirements.txt
Create a .env file and add GOOGLE_AI_KEY and DISCORD_BOT_TOKEN in as specified in .env.development file
Run as python main.py
Optionally, you can configure the bot as follows using settings.py:
Add custom initial conversation for every chat by editing the BOT_TEMPLATE variable as follows:
BOT_TEMPLATE = [
{'role':'user','parts': ["Hi!"]},
{'role':'model','parts': ["Hello! I am a Discord bot!"]},
{'role':'user','parts': ["Please give short and concise answers!"]},
{'role':'model','parts': ["I will try my best!"]},
]
Change content safety settings as follows:
SAFETY_SETTINGS = [
{"category": "HARM_CATEGORY_HARASSMENT", "threshold": "BLOCK_NONE"},
{"category": "HARM_CATEGORY_HATE_SPEECH", "threshold": "BLOCK_NONE"},
{"category": "HARM_CATEGORY_SEXUALLY_EXPLICIT", "threshold": "BLOCK_NONE"},
{"category": "HARM_CATEGORY_DANGEROUS_CONTENT", "threshold": "BLOCK_NONE"}
]
Change AI generation parameters using the variables TEXT_GENERATION_CONFIG and IMAGE_GENERATION_CONFIG
Error logs are stored in the errors.log file created at runtime.
Chat data is stored between bot runs using shelve.