Betit is a Claude Code plugin that turns every task into a wager. Before writing a line of code, Claude bets 5–50 points — sized to its real confidence — that it ships your task correctly in one shot. You're the only judge: any bug, correction, or second attempt is a loss. Wins pay the wager, three-win streaks earn bonuses, and the bank persists across every session and project. What's different: AI assistants always sound confident. Betit makes that confidence cost something.
No reviews yetBe the first to leave a review for Betit
Maker
📌
Betit is a Claude Code plugin that turns every task into a wager. Before writing any code, Claude bets 5–50 points that it ships your task correctly in one shot. You're the only judge — any bug, correction, or second attempt is a loss. Wins pay the wager, streaks earn bonuses, and the bank persists across every session. AI assistants always sound confident; betit makes that confidence cost something, so Claude learns to size its bets honestly instead of bluffing.
Report
Love the skin-in-the-game angle here, the confidence-as-wager mechanic is clever and the persistent bank sounds satisfying to watch grow. One idea from a user who's tried a lot of these plugins: add a session history view that shows wager size, outcome, and the actual diff or task description for each bet, so I can audit where Claude's confidence is well-calibrated versus where it's bluffing. Would make the whole loop way more educational and less of a black box.
Report
Love the concept of skin in the game, this is genuinely clever. One thing I'd love to see: a small history view showing past wagers and outcomes per project, so you can spot which task types Claude keeps overconfident on and adjust your prompting accordingly.
Report
honestly the confidence-to-wager mapping is such a clever mechanic, basically forcing the model to quantify its own uncertainty instead of just bullshitting you. love that the bank carries across sessions too, makes it feel like an actual ongoing thing.
Report
finally something that makes my AI assistant put its confidence where its mouth is, love watching the streak bonuses rack up when it actually nails something first try
Love the skin-in-the-game angle here, the confidence-as-wager mechanic is clever and the persistent bank sounds satisfying to watch grow. One idea from a user who's tried a lot of these plugins: add a session history view that shows wager size, outcome, and the actual diff or task description for each bet, so I can audit where Claude's confidence is well-calibrated versus where it's bluffing. Would make the whole loop way more educational and less of a black box.
Love the concept of skin in the game, this is genuinely clever. One thing I'd love to see: a small history view showing past wagers and outcomes per project, so you can spot which task types Claude keeps overconfident on and adjust your prompting accordingly.
honestly the confidence-to-wager mapping is such a clever mechanic, basically forcing the model to quantify its own uncertainty instead of just bullshitting you. love that the bank carries across sessions too, makes it feel like an actual ongoing thing.
finally something that makes my AI assistant put its confidence where its mouth is, love watching the streak bonuses rack up when it actually nails something first try