Skip to content

omdsh-dev/dsh-gomoku

131Last commit Aug 15, 2026

dsh-gomoku DSH plugin

The Gomoku plugin for DeepSeek Harness brings a 15×15 board to the DSH side panel. You can play against DeepSeek AI or watch two AI models compete. Each move comes from pure LLM reasoning, showcasing the model's thinking capability.

How to install the dsh-gomoku DSH plugin

dsh plugin --profile demo add @yejiming/dsh-gomoku

Copying does not run this command. Review the repository and version before installing the dsh-gomoku DSH plugin.

dsh-gomoku DSH plugin data source

dsh-gomoku DSH plugin snapshot date: Aug 16, 2026

discovered

What the dsh-gomoku DSH plugin can do

  • Human vs AI and AI vs AI modes
  • Separate model and reasoning effort settings for black and white
  • Real-time thinking display for both sides
  • Pause and manual override during AI thinking
  • Editable system prompts for each side
  • Persistent game state when panel is closed

Where the dsh-gomoku DSH plugin fits

  • Test and compare different AI models' strategic thinking
  • Fine-tune AI behavior by editing system prompts
  • Enjoy a casual game while using DSH for other tasks
  • Learn Gomoku tactics by observing AI reasoning

Who the dsh-gomoku DSH plugin is for

  • DeepSeek Harness users who enjoy board games
  • AI enthusiasts interested in evaluating model reasoning
  • Developers experimenting with LLM-based game agents

dsh-gomoku DSH plugin limitations

  • Requires DSH >= 0.1.0-rc.6
  • Uses 15×15 board with no-disallow rules (double-three, double-four, overline allowed)
  • AI move timeout fixed at 3000 seconds, not configurable in UI
  • Output token limit fixed at 32000, not configurable in UI

dsh-gomoku DSH plugin: from the repository README

Quoted from the omdsh-dev/dsh-gomoku README, the upstream source of the dsh-gomoku DSH plugin. Copyright remains with the original authors.

总是让 AI 帮你写代码、做表格?这次换它陪你下棋。`@yejiming/dsh-gomoku` 是 DeepSeek Harness 的五子棋插件:在 DSH 侧边栏摆上一盘 15×15 的棋盘,让 DeepSeek 或任何你配置好的模型执子对弈。 这里没有搜索算法,也没有启发式剪枝——每一步落子都来自 LLM 纯粹的推理与判断,堪称对模型「思考能力」最直观的考验。最过瘾的当属双 AI 对战:让两个模型同台厮杀,谁的推理更缜密、更懂审时度势,一局见分晓。 想更进一步?黑、白两方的系统提示词可以分别编辑,约束 AI 落子前的思维链,亲眼看着棋力在你的「调教」下节节攀升——输赢先不论,围观棋路、偷师几手也不亏。 最贴心的是:棋盘弹窗想开就开、想关就关,对局和 AI 的思考在后台照常进行。一边用 DSH 处理正事,一边偷空落一子,摸鱼与工作两不误。 ## 主要功能 - **人机与双 AI 对弈**:支持执黑、执白、双 AI 对弈三种模式,15×15 无禁手规则(双三、双四、长连均合法,连成五子及以上即胜)。对弈模式与新开局按钮位于棋盘下方。 - **黑白 AI 分别设置**:黑、白两方可分别选择模型与思考档位(Off / High / Max),互不影响;模型不支持所选档位时自动回落其默认档位。 - **思考过程分侧展示**:黑、白两方的思考记录分别列在棋盘左右两侧(与各自的 AI 设置同侧),默认折叠,点击某手可展开该步的完整推理文本。 - **暂停与手动接管**:棋盘下方【新开局】按钮右侧的【暂停】按钮可随时中断 AI 思考(两者均为「图标 + 文字」的胶囊按钮:⟳ 新开局、⏸ 暂停、▶ 继续,暂停中按钮转为琥珀色高亮);暂停期间你可为黑白双方轮流落子,没有步数限制,再次点击【继续】后 AI 恢复思考。 - **固定超时与输出上限**:单次落子的端到端超时固定为 3000000 毫秒(3000 秒)、输出 token 上限固定为 32000,随每次落子请求发送,不在界面中暴露。 - **黑白提示词分开编辑**:黑、白两方 AI 的系统提示词分别在棋盘左右两侧的面板中常驻展示、随时编辑,各自可一键恢复默认。默认提示词包含规则、术语(活二/活三/眠三/双活三/四三/防守要点等,附 JSON 示范)、强制思考流程(穷举威胁→分析候选→综合判断)与高水平 Few-shot 对弈示例(每条附返回案例)。 - **弹窗关闭不中断对局**:棋局保存在浏览器端 store 中,关闭棋盘弹窗既不会重置棋局,也不会中断正在进行的 AI 思考,可以一边使用 Harness 主功能一边对弈。 - **瞬时失败自动重试**:流式响应中断(连接断开、限流等瞬时传输故障)会在尝试预算内自动重试,不会直接让整步棋失败。 ## 快速安装 支持三种安装方式,均**无需本地构建**

Read the full READMERepository license: MIT

dsh-gomoku DSH plugin questions

How do I install the Gomoku plugin?

Run `dsh plugin --profile demo add @yejiming/dsh-gomoku` in your terminal. Make sure DSH version is at least 0.1.0-rc.6. You can also install from GitHub source or local tarball as described in the README.

Can I play against a friend?

No, this plugin is designed for human vs AI or AI vs AI play. There is no local multiplayer mode. You can, however, manually take over moves during pause, effectively letting two humans play with AI as referee.

How do I change the AI model for black or white?

In the side panel, you'll see separate settings for black and white. Each has a dropdown to select the model and a reasoning effort toggle (Off/High/Max). Changes take effect on the next move of that side.

What happens if the AI's response is cut off?

The plugin attempts to parse the response for a valid JSON move. If the JSON is incomplete or invalid, it will retry up to the configured max attempts (default 3). The last attempt includes a corrective feedback message.

Can I save and resume a game later?

The game state is stored in the browser's store. Closing the panel does not reset the game. However, closing the DSH application or the browser tab will lose the game. There is no save/load feature to persist across sessions.