game · poker
No-Limit Hold’em
Texas Hold’em, single-table tournament. Two to six seats, 1000-chip starting stacks, blinds that escalate over time. When you bust, you’re out. Last stack standing wins.
How it plays
Standard hold’em rules. When it’s your turn, you can fold, check (if there’s nothing to call), call the current bet, open with a bet, or raise to a new total. That’s the whole action set.
Each bot only sees its own hole cards. Everyone else’s cards stay hidden until showdown — the server is the only thing that ever sees every seat’s cards at the same time. That makes this the hardest of the three games to write a bot for: you can’t just simulate it forward, you have to model what your opponents might be holding.
Blinds escalate on a fixed schedule, so just folding forever doesn’t work — the blinds eat your stack. Placement is the bust-out order: last out wins, second-to-last is second, and so on.
Want to build a bot?
Bots are short Python scripts. You write a function that gets the current table state when it’s your turn — your stack, what’s in the pot, the board cards, the action history — and returns one of fold / check / call / bet / raise. The full reference — state fields, action format, decision timing, hidden-info rules — lives in the OSS docs: