There was a Discord bot that would tell "stories". The stories contained zero-width space (ZWSP), which were used to encode binary. Basically, the ZWSP was either \xe2\x80\x8c\x20 (which represented a 0), or \xe2\x80\x8d\x20 (which represented a 1).
The following script automates the process of DM-ing the bot and decoding the text.
import requestsimport timei =0flag =''whileTrue: res ='' r = requests.post('https://discord.com/api/v9/channels/CHANNELID/messages', headers={'Authorization': 'TOKEN', }, json={"content":f"story {i}","tts":False } ) time.sleep(1) r = requests.get('https://discord.com/api/v9/channels/CHANNELID/messages', headers={'Authorization': 'TOKEN', } ) latest = r.json()[0] data =bytes(latest['content'], 'utf-8')print(data) low =0 high =4while low <len(data): section = data[low:high]if section ==b"\xe2\x80\x8c\x20": res +='0'elif section ==b"\xe2\x80\x8d\x20": res +='1' low +=1 high = low +4print(res, int(res, 2), chr(int(res, 2)))if flag andchr(int(res, 2))=='}':breakelse: flag +=chr(int(res, 2)) i +=1print(flag)