Skip to content

A Hanabi bot written from scratch in NodeJS, named after WillFlame's Will-Bot

Notifications You must be signed in to change notification settings

IceSelkie/billbot-baggins

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

36 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

BillBot Baggins

Billbot is a Hanabi bot written from scratch in NodeJS. It is named after WillFlame14's bot named "will-bot".

Its goal is to use Hatguessing to maximize shared information, while remaining very fast. The original algorithm I am implementing is from the 2014 paper "How to Make the Perfect Fireworks Display: Two Strategies for Hanabi", where Cox et al. proposed a simple hatguessing algorithm to decide on actions and provide hatguessing hints and hint interpretations. More involved hatguessing algorithms exist which can work almost magically even on difficult variants, such as Yagami_Black's bot (who also has a modified version that is playable by humans that is documented here).

Running the bot

This is currently a work-in-progress and currently will not function as-is. You may need to create and/or move some directories to get it to function. Create an account at hanab.live (generously hosted by Zamiel), and save the token to the file "token_v2.json" with the format: {"username":{"token":["hanabi.sid=BASE64TOKENGOESHERE="]}} Install NodeJS and use npm i to install the project's dependencies (only the library ws) Build the derived data with node 0_run_all.js for the bot from the variant descriptions, then finally: Start the bot from the command line with node 0_start_bot.js.

Using the Bot

Similar to WillFlame's bot, pm the bot /join to have it join your table, or /join {password} if you have a password on the room. Pm the bot /leave to make it leave, or /leaveall in the table's chat to make all bots leave.

About

A Hanabi bot written from scratch in NodeJS, named after WillFlame's Will-Bot

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published