dsh-gomoku: Play Gomoku with AI in DeepSeek Harness

omdsh-dev/dsh-gomoku

EntertainmentVerified
Listed on 2026-08-20
Page last updated 2026-08-20

Play Gomoku with AI in the DeepSeek Harness sidebar, supporting human-vs-AI and AI-vs-AI matches to test model reasoning.

Turn the DeepSeek Harness sidebar into a board to play against AI or pit two AIs against each other, directly testing model reasoning.

install
dsh plugin --profile web add github:omdsh-dev/dsh-gomoku
Category
Entertainment
Platform
DSH-Plugin
Author
omdsh-dev
Distribution
Plugin

dsh-gomoku Key Features

Human vs AI & AI vs AISeparate black/white AI configSide-by-side reasoning displayPause and manual takeoverEditable prompts

dsh-gomoku Repository

omdsh-dev
Publisher
omdsh-dev
Repository
omdsh-dev/dsh-gomoku
14
Stars
1
Forks
0
Watchers
1
Open issues
LanguageTypeScript
LicenseMIT License
Website
Repo last push2026-08-15 13:42:26
Repo created2026-08-09 01:27:52
Topics
deepseek-harnessdshdsh-plugin
Default branchmain

dsh-gomoku Repo Summary

dsh-gomoku is an entertainment DSH plugin for DeepSeek Harness that lets you play Gomoku against an AI in the DSH sidebar, or pit two AIs against each other to see which has stronger play. It addresses the need to have AI play with you—no search algorithms, every move comes from pure LLM reasoning, directly testing the model's thinking ability. Core features include human-vs-AI and AI-vs-AI modes, separate model and reasoning effort settings for black and white, side-by-side reasoning display, pause and manual takeover, editable system prompts for each side, and game persistence when the popup is closed. The plugin is maintained by omdsh-dev, licensed under MIT, written in TypeScript, and last updated in 2026-08.

What game modes does the dsh-gomoku DSH plugin support?

It supports three modes: play as black, play as white, and AI vs AI, on a 15×15 board with freestyle rules (no forbidden moves—double-three, double-four, and overlines are legal; five or more in a row wins). Black and white can each select a model and reasoning effort (Off / High / Max) independently, falling back to the model's default if unsupported. Reasoning logs are shown on the left and right sides of the board, collapsed by default; clicking a move expands the full reasoning text.

How do I install dsh-gomoku?

Three installation methods are supported, none requiring local build (the lib/ build output is committed). npm installation is recommended, but you can also install from GitHub source or a local tarball. Requires dsh ≥ 0.1.0-rc.6. The installation command is:

bash
dsh plugin --profile web add github:omdsh-dev/dsh-gomoku

After installation, verify the layer with --dump-config, then start:

bash
dsh --profile demo --dump-config
dsh --profile demo

What are the AI move request and configuration options for dsh-gomoku?

Each AI move is an independent auxiliary request to the selected provider/model route. System prompts are editable per side, with defaults including rules, terminology, a forced thinking process, and few-shot examples. Configuration keys include moveTimeoutMs (default 3000000 ms), maxMoveOutputTokens (default 32000), and maxMoveAttempts (default 3), sent fixed by the browser and not exposed in the UI. Transient stream failures are retried within the attempt budget.

Which DeepSeek Harness versions and platforms are compatible with dsh-gomoku?

The plugin uses the webServer service from @deepseek-ai/dsh-host-webserver, requiring dsh ≥ 0.1.0-rc.6; earlier versions lack this service, causing the plugin to stay pending. The game state is stored in the browser store, so closing the popup does not interrupt the game or AI thinking, allowing you to play while using Harness's main features.

View more DSH plugins at dsh-plugin.org

DSH-Plugin FAQ