尚未检测到更新。GitHub 发布版本后会显示在这里。
个人 GitHub 账号下公开仓库的所有者可以登录并验证所有权,自行管理介绍,无需管理员批准。
登录并生成验证文件,在 30 分钟内将其提交到此仓库的默认分支。其他分支或分叉仓库中的文件无效。然后在下方检查文件,即可无需管理员批准直接管理项目。请保留文件以维持管理权限。
如需登录,将前往登录页面。登录后请回到此处,再次点击所选的验证按钮。
在项目文件夹中运行 /nomorevibe verify,即可验证所有权并自行管理介绍。
一分钟完成验证制作者的介绍仅有上面的一句话。 更多详情请阅读仓库 README。 仓库 README ↗
正在加载完整 README。
查看完整 README ↗Double-six partner dominoes (4 players, 2 teams). You + an AI partner vs two AI opponents. Everyone draws 7 — the full 28-tile set, no boneyard. First team to 100 points wins.
The table is square (1:1). Tiles are drawn at a fixed size and never shrink — the chain starts in the middle and, when an arm reaches the edge, it turns the corner and runs along the border, coiling inward if it goes all the way around.
A full 28-tile chain is 1372px of dominoes and needs a 580px square to coil cleanly. Smaller screens lay it out at that size and scale the whole table down, so tile size is still fixed for the entire round.
Three difficulty levels, selectable during play.
The AI opens using the classic Venezuelan school — the regla de oro of El Tigre de Carayaca (Héctor Simosa Alarcón): lead the highest double that has company, fall back to the highest bare double, and with no doubles lead a heavy tile from your longest suit.
Measured over 600 mirrored matches (sides swapped so seat order cannot skew it):
| Matchup | Win rate |
|---|---|
| hard vs medium | 65.8% |
| hard vs easy | 81.3% |
| medium vs easy | 80.7% |
| hard vs hard | 50.7% (sanity check) |
After every round you can open a review of your own play. This is a team game, so the review is built on the partner doctrine of Héctor Simosa Alarcón, "El Tigre de Carayaca" (Ciencia y Arte en el Dominó). Two independent lenses:
Principles. Judged against the classic partner ideas, with a separate team-play score for how often your choices helped your side:
Engine analysis. Separately, where your move ranked in the engine's own search, what it would have played instead, and why.
Forced moves are never counted against you.
npm install
npm run dev # http://localhost:3000
npm test # engine unit tests
npm i -g vercel
vercel # log in with your account, accept defaults
Or push the repo to GitHub and import it at vercel.com/new.
src/engine/ — pure game logic (tiles, rules, scoring) + AI. No UI dependencies, so it can run server-side for online multiplayer in phase 2.src/components/ — the game table UI.src/app/ — Next.js app shell.Private tables with a shareable six-character code. Create one, send the code to three friends, and empty seats can be filled by the computer (host's choice, with a skill setting).
Turn on the backend by following SUPABASE.md — one free project and three environment variables.
How it is kept honest. The browser never runs the rules and never holds the deck. It sends intentions ("play this tile on that end") to the server, which owns the authoritative game state, validates against the same engine the solo game uses, and replies with a view containing only that player's own tiles. Everyone else's hand is reduced to a count. Row level security denies browsers all direct database access, so hidden information stays hidden even if someone goes looking with the developer tools.
Other details:
localStorage is your seat.
The schema already carries a user_id column, so real accounts and ratings can
be added later without a rewrite.