Play and learn Texas Hold'em poker in Claude Code
How it works
You play against three AI opponents (Alex, Jordan, Sam), each running as a separate Claude subagent. Each opponent runs in its own isolated context and can only see its own cards, so no one has access to anyone else's hand. The main agent acts as dealer, game runner, and optionally your coach.
What you get
- Full No-Limit Texas Hold'em cash game (4 players, 1000 chips, 5/10 blinds)
- Three opponents with distinct play styles you'll need to figure out over time
- ASCII table visualization updated before every decision
- Three coaching modes: no coaching, hints + post-hand analysis, or post-hand analysis only
- Mix of random and crafted hands designed to teach specific concepts (pot odds, position, kicker awareness, knowing when to fold)
- Shorthand input:
f,c,r 90,allin
Setup
curl -sL https://raw.githubusercontent.com/BohdanPetryshyn/code-royale/main/SKILL.md \
-o ~/.claude/skills/code-royale/SKILL.md --create-dirsThen start Claude Code and say something like "let's play poker" or "deal me in".
Model recommendations
- Opus recommended for the main agent (dealer/coach) for best coaching and game flow
- Sonnet also works well as the main agent
- Opponents always use Sonnet subagents regardless of your main model
License
MIT