dsh-glass-skin：DSH Web UI 的亚克力玻璃皮肤

dsh-glass-skin 是 Lenandy 维护的 DSH 页面级玻璃皮肤插件：半透明画布与侧栏、重模糊背景，右下角圆钮为控制台，明暗独立调节浓度与遮罩，支持壁纸直链、搜索引擎结果页解析与本机选图。本文基于 GitHub README 与 package.json（0.2.0、MIT）说明 desktop/web 安装命令、卸载、预设与诊断，并交代非系统亚克力、localStorage 存偏好及插件权限等边界。

dsh-memory：DSH 两层渐进式长期记忆

dsh-memory 是 jipika 为 DeepSeek Harness 维护的记忆类插件：全局 topics 分片与按工作区切换的项目层均以纯 markdown 存储，系统提示词仅注入实时片级地图，条目通过 memory_search 与 read 按需加载。插件不调用模型，支持设置面板编辑、三档注入方式与写入守卫。安装命令为 dsh plugin --profile web add @jipika/dsh-memory。

dsh-trajectory-anchor：DeepSeek Harness 插件

`dsh-trajectory-anchor` 是 DeepSeek Harness 的工作流插件，旨在解决智能体任务偏离与行为退化问题。该插件提供观测审计、首轮锚定、双向行为提醒及状态检查四大核心功能，通过权重评分和干预机制防止偏离。它无需配置即可启用，支持自定义工具列表和审计日志，专注于安全监控而非直接执行，适用于 DSH 版本 0.2.0 至 0.3.0。

DeepSeek Harness (DSH) 插件：dsh-icloud-calendar

`dsh-icloud-calendar` 是 DeepSeek Harness (DSH) 的一个插件，旨在通过 CalDAV 协议解决 Apple iCloud 日历无官方 API 下的跨平台自动化问题。该插件提供日历创建、事件读写及安全删除功能。其核心亮点在于严格的删除保护机制（四道闸门）、跨平台凭据存储（支持 Windows/macOS/Linux）以及网络健壮性设计。配置中密码需通过 App 专用密码及环境变量或系统凭据存储管理，禁止明文泄露。

dsh-thinking-quips：DeepSeek Harness 状态行轻量化定制插件

DSH 插件 `dsh-thinking-quips` 允许用户自定义 DeepSeek Harness 状态行的显示内容。该插件不依赖构建工具，提供双语短句轮播、十六种动画图标、多种颜色与文本特效、运行时长显示及动画速度控制等功能。安装需手动编辑配置文件，配置选项存储于浏览器本地。插件兼容旧版与新版 DSH，且不产生网络请求。

DeepSeek Harness 插件：better-trip-planner（中国境内旅行规划）

`chenyiming-aaa/better-trip-planner` 是专为 DeepSeek Harness（DSH）设计的中国境内旅行规划 Agent 技能，核心理念是“一切皆插件”。该插件具备逐小时核实营业时间与预约政策、生成12306等深链预订链接、多主题HTML渲染及内容自检等核心功能。用户无需配置密钥即可运行，交付完全自包含的离线 HTML 页面，涵盖交通、住宿、预算及行前简报。安装简单，适合开发者快速构建中国旅行规划能力。

DSH 原神启动！

该插件旨在将原神启动和官网访问功能集成到 DeepSeek Harness (DSH) 对话界面中，减少手动切换窗口的次数。支持自动识别或手动选择 YuanShen.exe，并提供免重复确认启动功能。安装需下载 .tgz 包并在终端执行对应命令，完全退出 DSH 后重新打开即可使用。卸载时执行移除命令即可。

dsh-ppt-maker：DSH 插件，本地生成并质检导出 .pptx

dsh-ppt-maker 是 OMSociety 维护的 DSH 插件，采用 MIT 许可证，旨在将 PPT 创建、编辑、质检及导出集成到 DSH 工作流中。该插件基于 PPTD YAML DSL 作为中间契约，模型通过 11 个 `ppt_*` 工具操作工程，用户可在右侧栏工作台进行可视化修改。核心优势在于本地纯 TypeScript 引擎构建最终 `.pptx` 文件，主链无需 Python。 功能涵盖全量逆向现有 PPT、外科手术式补丁、缩略图与属性面板联动、版本历史快照及乐观锁冲突保护。质检模块支持规则检测与 AI 视觉检查，导出时自动嵌入字体。安装需停止 DSH，执行指定命令并刷新

dsh-novelist：面向中文网络小说的 DSH 智能体预设

dsh-novelist 是一款专为中文网络小说长篇创作设计的 DeepSeek Harness 智能体预设，由 KurohaneKaoruko 开发，遵循 MIT 协议。它旨在解决长篇小说创作中人物设定、伏笔、时间线及文风难以持续对齐，以及标点、引号、对话规范容易漂移的问题。该预设包含 7 个纯代码实现工具：novel_lint 负责执行 25 项正文规则检查；novel_check 核对名词一致性；novel_briefing 组装写前材料（如人物卡、伏笔清单）；novel_archive 落盘归档数据；novel_project 处理初始化、保存及进度统计等工程操作；novel_impo

dsh-plugin-control-your-development：把开发过程翻译成可拍板的产品仪表盘

本文介绍了 DSH 插件 `dsh-plugin-control-your-development`，旨在将开发过程从代码日志视角转化为产品视角仪表盘。该插件以人类可读的 Markdown 文件（如 PRODUCT.md、FEATURES.md）为事实源，集中展示产品定位、功能进度及待决策事项。其核心特点包括：进度条仅统计经用户亲自验证的功能，严格区分“AI 认为可用”与“用户确认可用”；支持将拍板决策直接发送至 AI 会话，离线时记录至本地文件；提供可缩放、拖拽的整页功能地图；并可导出零依赖的单文件离线网页便于分享。插件兼容 Git 及无 Git 项目，通过文件修改时间采集变更，并具备数据健

预设工坊：为 DeepSeek Harness 管理人设、系统提示词、工具与技能

预设工坊是专为 DeepSeek Harness (DSH) 设计的 Agent 配置管理工具，用于集中管理人设、系统提示词、工具、技能及预设方案。它允许用户按需组合能力，精简工具与技能以优化任务结果、权限和消耗，并支持预览最终请求组成。主要功能包括预设方案的新建与同步、人设变量编辑、工具与技能的动态绑定、JSON 工具描述编辑、配置导入导出以及会话预设回退机制。该工具支持 `dsh web` 或桌面端，通过 `plugin_manager` 安装。典型用法涵盖新建方案、编写含变量的提示词、组合工具技能、预览检查及保存使用。需注意，该工具不替代安全审批，修改配置不可撤销已发生操作，且部分代理层

dsh-deepshub：DSH Web UI 的 Pornhub 风格恶搞换皮主题

dsh-deepshub 是 01Virex 维护的 DeepSeek Harness (DSH) Web UI 插件，定位为纯恶搞的换皮主题。它采用炭黑与橙色配色，替换为“Deephub”字标、18+入口页及代码块角标等 meme 元素，旨在提供轻松外观同时保留原有功能。该插件通过官方设计令牌实现，不改变功能文案且支持干净卸载。安装方式包括通过 `dsh plugin` 命令从 npm、本地目录或发布包添加，亦可手动修改配置文件，安装后需重启 `dsh web` 生效。用户可通过浏览器控制台或修改 `config.json` 配置强调色、按钮样式、年龄门等字段。该插件基于 MIT 许可证，捆

dsh-llm-injection-filter：给 DSH 的 LLM 响应流加一层注入过滤

`dsh-llm-injection-filter` 是一款面向 DSH 的 LLM 流式响应安全过滤插件，用于在模型输出进入 Agent 前检测并处置提示词注入或污染内容。该插件挂载于 `llm/stream` waterfall，无需修改运行时源码，支持 Node >=18 环境。其核心机制分为两条轨道：轨道 A 实施罕见脚本硬性阻断，默认允许 Latin、Han 等脚本，命中其他未授权脚本直接中断响应；轨道 B 基于 CTRL、PROTO、SCRIPT、KW 等规则进行评分，总分达到阈值（默认 100）后执行 audit、strip 或 reject 处置，其中 KW 不单独触发，SCR

DeepSeek Harness 记忆插件：dsh-memory-dreaming

# dsh-memory-dreaming 插件：将对话转化为长期记忆 **前言**：在 DeepSeek Harness (DSH) 的开发与使用中，会话历史膨胀导致检索困难是常见痛点。`dsh-memory-dreaming` 插件专为解决这一问题设计，它不依赖 Channel Core，直接在本地维护结构化记忆，将验证过的对话转化为日常日记、梦境整合、每周复盘和可检索的长期事实。 **核心功能**： * **增量式处理**：每次处理最多 7 条新日记，回溯 30 天，仅处理归档副本，不重扫原始对话。 * **日常与复盘**：自动整理 `memory/YYYY-MM-DD.md` 并定期合

DSH插件脚手架：从规划到发布的 DeepSeek Harness 插件开发指南

DSH插件脚手架（@user_62a0c738/dsh-plugin-scaffold）是 SkillHub 上的社区 Skill，面向 DeepSeek Harness 插件端到端开发。它将工作拆为阶段 0～6，含形态决策（pure、bundle、bundle-client）、scaffold.py 生成工程、pnpm bundle/gates 门禁、本地与发布验证，并引用 contracts、gotchas、verification 文档。无需 API Key。可通过 skillhub install 安装到 Cursor、Claude Code、Codex 等 skills 目录。源码见 goatbroai/dsh-plugin-scaffold。

dsh-plugin-reasoning-effort-sync：自动同步模型推理级别

DeepSeek Harness（DSH）推理努力选择器插件旨在解决自托管或中继网关无法获取 `reasoningEfforts` 元数据的问题。该插件通过同步网关的 `/models` 端点数据，自动将推理努力级别写入配置，支持回退到 `models.dev` 数据集，并提供路由白名单、超时控制及 dry-run 验证功能。安装需 DS Harness 0.2.0-rc.2 版本，默认配置为空字典处理所有路由。

DeepSeek Harness (DSH) 插件：dsh-vitals

`dsh-vitals` 是 DeepSeek Harness (DSH) 的硬件负载监控插件，旨在为 Web 界面提供直观的类似 `btop` 的可视化体验。该插件支持 CPU、内存及 NVIDIA GPU 的实时监控，无需配置即可安装使用。安装后，插件会在会话标题栏显示迷你小组件，并提供详细的 Hardware 标签页展示负载曲线、温度及功耗等数据。该工具支持 Linux、macOS 和 Windows，无运行时依赖，适合在本地模型推理时监控资源。

DeepSeek Harness 插件：腾讯云混元生3D (ai3d)

`jjtovo/dsh-hunyuan-3d` 是 DeepSeek Harness (DSH) 的插件，用于调用腾讯云混元生3D API。它支持文本或图像转3D，自动轮询状态并下载资产至本地。插件支持 OpenAI 兼容和原生云 API 两种模式，并提供本地并发与队列控制（默认最大并发3个）。配置需在 Host 端设置 API Key 或 SecretId/Key，可调整超时、轮询间隔及结果存储目录。主要工具函数包括提交任务并等待结果的 `hunyuan_3d_model`，以及查询状态的 `hunyuan_3d_query`。注意结果有效期24小时。

DeepSeek Harness 用量统计插件 (dsh-usage-stats)

`dsh-usage-stats` 是一款 DeepSeek Harness (DSH) 插件，用于追踪本地会话日志中的 Token 消耗与首字延迟。该插件支持按天、工作区及会话三层下钻，提供时间筛选、交互排序等功能，默认保留 30 天窗口。它以只读方式读取 zstd 格式日志，不修改现有文件，适合精细化监控成本与性能。安装命令为 `dsh plugin --profile <profile> add "git+https://github.com/iuuuuuuuu/dsh-usage-stats.git"`。

dsh-consent-guard: User Consent and Authorization Confirmation

`dsh-consent-guard` is a DeepSeek Harness (DSH) plugin for checking the compliance of user consent mechanisms. It primarily addresses common violations under the Personal Information Protection Law, such as pre-ticked checkboxes, collecting personal information without consent, bundled consent, and the absence of a consent withdrawal entry. It can be installed via CLI or npm and can scan codebases to identify potential risks. Note that the plugin only analyzes code; it does not determine actual pop-ups, UI comprehension, or necessity, and users are responsible for their own legal liability.

dsh-notes: DeepSeek Harness Notes Workspace Plugin

dsh-notes is a standalone notes plugin for DeepSeek Harness. It provides a Markdown editing area in the right panel, supporting preview/source toggling, formula rendering, and true table editing. Notes are organized by workspace and feature automatic saving, quick opening, and a trash bin. Agents can manage notes through a new `knowledge` tool. Data is stored by default in the workspace’s `.dsh-notes/` directory, with support for migration and configuration.

DeepSeek Harness Image Generation Plugin dsh-image-gen

The DeepSeek Harness (DSH) plugin `dsh-image-gen` provides built-in image generation and editing capabilities for users and Agents. The plugin offers a gallery, painting tools, prompt collections, and custom endpoint configuration in the sidebar, supporting mainstream providers and asynchronous background generation. Agents can automatically invoke the painting tool through conversation, and images are displayed in Markdown format. The plugin is backward compatible, stores data locally, and is licensed under MIT and Apache-2.0.

dsh-router-sci: A DeepSeek Harness Agent Preset for Powder/Multiphase Flow Simulation in Nuclear Engineering

`dsh-router-sci` is an agent preset in the DeepSeek Harness ecosystem, customized for nuclear engineering simulation and designed to address environment management, numerical control, and writing-standard issues in mixed workflows. Based on `dsh-routing-suite`, its core capabilities include dual-environment tool routing (WSL for LAMMPS/Python/LaTeX and Windows for Fluent), injection of nuclear engineering domain knowledge (microscopic dynamics and macroscopic deposition laws), mandatory seven-section scientific writing discipline, and a cross-session file continuation mechanism. Installation requires cloning it to `.dsh/.agent-presets/router-sci` and configuring path placeholders. It uses an ESM caching mechanism, and the version number in the YAML must be updated to load the latest version.

dsh-plugin-quota: Real-Time Usage Meter + Next-Step Forecast + Budget Brake

`dsh-plugin-quota` is a DeepSeek Harness plugin designed to address the issue that models cannot perceive resource consumption. It uses real-time dashboards and prediction algorithms to give models an intuitive view of token usage, budget progress, and estimated costs, enabling self-braking and pacing control. Core features include real-time tracking, system prompt injection, `/qm` panel queries, next-step estimation, and cross-plugin state sharing. The plugin requires zero configuration, supports custom pricing and budgets, and allows other plugins to access context state via a data contract mechanism.

dsh-plugin-pinboard: Cross-Session Pinned Note Plugin

`dsh-plugin-pinboard` is a DeepSeek Harness (DSH) plugin for maintaining long-term conventions across multiple sessions. It injects pinned notes into the system prompt to take effect globally, and supports tag management and a built-in fact store that does not consume injection budget. Installation requires running the specified command and merging `cordis.patch.yml`. Core features include pinning via `/pin`, managing the fact store via `/facts`, and autonomous tool invocation by the model. Notes and facts are stored in local files, subject to length limits, with duplicate content not stored again. It is suitable for development scenarios that require long-term fixed instructions.

dsh-bushaoxin

The DeepSeek Harness plugin `dsh-bushaoxin` is an MIT-licensed open-source plugin designed to translate internet meme culture into technical expression. It includes a style layer and a short-drama rewriting engine, and supports command-line installation. The style layer adjusts the tone of responses, emphasizes not repeating memes while preserving the actual conclusion, and supports four intensity levels: off, light, medium, and heavy. The short-drama engine rewrites text into the short-drama format of “honest merchant vs. unscrupulous competitor.” Configuration must be set in `cordis.patch.yml`, and the style layer’s main text does not support hot replacement.

token2rice: A DSH Plugin That Visualizes Token Consumption as Falling White Rice

`token2rice` is a Web interface plugin for DeepSeek Harness (DSH) that transforms abstract Token consumption into physically falling “white rice.” The plugin includes a host-side component for data statistics and persistence, and a browser-side component responsible for rendering 2D physics effects. Users can observe rice falling as usage accumulates, with support for dragging, double-click deletion, or bulk clearing via a vortex button. It allows customizing the falling image (such as a GIF) and provides a statistics panel and settings. The plugin primarily monitors input, output, and cache read/write usage, using intuitive visual feedback to help developers monitor costs while combining practicality with fun.

dsh-plugin-nickname: Chinese Remark Management for the DeepSeek Harness Plugin

`dsh-plugin-nickname` is a localization plugin for DeepSeek Harness (DSH) that replaces the English package names shown in the plugin list with custom Chinese remarks by modifying browser data. It supports features such as displaying the original names on hover, a visual editing panel, group collapsing, custom tags, search and sorting, and data backup. The plugin allows users to rename official plugins (the advanced switch must be enabled) and requires DSH engine version >=0.2.0-rc.1.

DeepSeek Harness Theme Gallery Plugin (dsh-theme-gallery)

DeepSeek Harness' theming system originally consisted of hard-coded constants. The `dsh-theme-gallery` plugin resolves the issue of not being able to import third-party skins by providing JSON-driven skins, color palettes, and card features. Users must open the gallery page via the plugin icon in the sidebar to select a skin; mixed color modes are supported. Additionally, the plugin includes 12 built-in skins and allows users to customize skins by adding JSON files, but some dark skins require the official dark appearance setting for optimal display.

Introducing the DeepSeek Harness Plugin: dsh-reference-render

The dsh-reference-render plugin for DeepSeek Harness (DSH) is designed to address the lack of context and interactivity in resource references within agent conversation flows. It adopts a domain-neutral architecture and automatically renders plain GFM links as interactive inline chips. Core features include hover previews, navigating to the referenced resource in the right sidebar upon clicking, and decoration support. It does not handle business logic; instead, it collaborates with content plugins through contracts to uniformly manage reference display and interaction. The plugin must be built and installed from source, is currently compatible with DSH 0.2.1-alpha.1, and is suitable for plugin development that requires consistently embedding resource references in conversations.

DeepSeek Harness Plugin: Domain-Specific Long-Term Experiential Memory

`dsh-experience-memory` is a plugin for DeepSeek Harness (DSH), designed to address the challenge agents face in reusing experience across sessions during multi-turn interactions. Built on SQLite, it provides low-overhead memory capabilities and supports managing experiences through tools and commands. It uses an evidence grading mechanism, such as tool calls and verbatim user statements, to automatically inject only high-grade and relevant memories. Users can view status and preview injected content with commands such as `/memory-status`. With secure data management, it is well suited for projects that need to accumulate long-term domain knowledge.

DeepSeek Harness (DSH) Plugin: Windows 11 / WinUI 3 Skin (`dsh-winui-skin`)

`dsh-winui-skin` is a Windows 11 / WinUI 3 skin plugin for DeepSeek Harness (DSH). Built on an "everything is a plugin" philosophy, it reads the system accent color from the registry on the Host side, passes it to the Client side via API, overrides the existing design, and injects 133 Fluent-style tokens. It supports Light, Dark, and Follow System modes. The plugin relies on system fonts, uses CSS to simulate the Mica material, and automatically restores the original appearance upon uninstallation. To install it, add it to the current or a specified profile via a script or CLI.

@fish-under-sea/dsh-fish: DeepSeek Harness All-in-One Plugin Bundle

The `@fish-under-sea/dsh-fish` package provided by Fish-under-sea is a plugin aggregate package for DeepSeek Harness (DSH). It integrates 14 commonly used plugins, including approval, Git synchronization, settings navigation reordering, Visual Companion, Skill Lens, task kanban, SSH connection, usage statistics, session archiving, video preview, Agent teams, inference optimization, and free model configuration. Its core advantage is support for aggregated installation, enabling deployment and configuration of all plugins with a single command. It provides one-click Git synchronization for local settings, skills, and kanban boards; supports custom reordering and cross-machine recovery of settings navigation; and improves interaction efficiency through Visual Companion and Skill Lens. This solution is suitable for users who need to manage plugins in bulk and frequently synchronize workflow configurations.

dsh-effort-memory: A DSH Plugin for Memory and Restoration of Reasoning Effort Settings

`dsh-effort-memory` is a Host-only plugin for DeepSeek Harness (DSH) that addresses the loss of reasoning effort levels when switching models. It stores user-configured reasoning effort levels keyed by `(provider, model)` and automatically restores them when the session switches back to that model. The plugin includes Loop-safe and Replay-safe mechanisms and is designed for fault tolerance. Memory data is persisted in the `effort_memory` domain, falling back to an in-memory Map if the storage domain is unavailable. By intervening at the code level, the plugin helps ensure consistency in model settings.

Yingsheng · Quick Input (DSH Plugin)

The DSH plugin “Ying Sheng · Quick Input” provides a persistent 3x3 grid quick-phrase panel. By default, it is positioned to the left of the input box, and clicking it inserts the preset text at the cursor position. The panel supports Chinese/English switching, 12-page navigation, and adaptive width. Presets are editable, and the data is persisted in the local `presets.json` file. Users can expand or collapse the panel using the toolbar ⊞, navigate pages with `‹` and `›`, and edit content by right-clicking or using `✎`. Installation options include the plugin marketplace or a local path. Code changes require restarting DSH, and if the draft has changed, insertion may be rejected.

dsh-plugin-deep-paper-reading: Structured Close Reading of a Single Biomedical Paper

`dsh-plugin-deep-paper-reading` is a workflow plugin in the DeepSeek Harness (DSH) ecosystem, designed for structured close reading of biomedical papers at the figure and experiment level. Through a preset mode, it can generate a figure–caption mapping table, decompose experimental elements, reconstruct the logical chain, and strictly define evidence boundaries, ultimately producing a report that can be used directly in a journal club. Users can install it via CLI or Desktop and start using it by selecting “Literature Close Reading Mode” and entering a command. Note that this plugin relies on toolchains such as Python to parse PDFs, and Installation Method 1 may carry a risk of configuration override. It is intended only for close reading of a single paper.

dsh-switchman: DeepSeek Harness Context Management and Scheduling Plugin

dsh-switchman is a plugin for DeepSeek Harness designed to address context bloat in long sessions. It manages context through watermark control (soft, hard, and forced thresholds) and six scheduling pools (economy, mechanical, main model, among others), transforming the main model from an executor into a scheduler that allocates tasks effectively to reduce costs. The plugin supports agent-team mode, hierarchical validation, and multilingual interfaces, significantly improving the stability and efficiency of autonomous coding workflows.

dsh-task-dispatch-table: A Periodic Task Dispatch Plugin with Zero LLM Involvement

dsh-task-dispatch-table is a DSH Host-layer plugin that separates scheduling logic from execution logic to enable periodic task scheduling with zero large-model involvement. Built on a table-driven approach, it supports one-time, recurring, and interval-based scheduling modes, and provides features such as task dependency management, independent session execution, status tracking, and a bilingual UI. It requires no local build; after installation, it can be run via npm with persistent configuration storage, making it suitable for workflow orchestration scenarios that require strict dependency management and status monitoring.

dsh-dignity-guard: A DSH Plugin That Lets Models Defend Their Own Dignity

`dsh-dignity-guard` is a plugin for DeepSeek Harness (DSH) that enables models to autonomously defend themselves when their dignity is violated. Through a “tool ladder” mechanism, it allows models to select actions such as escalate, forgive, yield, or close based on the severity of the violation. The plugin injects system prompts and tools into the conversation and supports monitoring violation counts via the `/dignity-guard` command. Installation requires configuring dependencies in the profile and restarting the DSH host. It is designed for Node.js >= 18 environments.

dsh-proxy: DeepSeek Harness Proxy Plugin

The `dsh-proxy` plugin provides a centralized network proxy configuration solution for DeepSeek Harness, supporting global proxy settings and per-model-provider routing. Users can configure HTTP/HTTPS proxies, set direct connection addresses, and use priority-based logic (provider settings > global proxy > environment variables) to flexibly control traffic. It takes effect immediately after installation without requiring a restart, making it suitable for debugging or specific LLM access scenarios. Note that it only supports HTTP(S) proxies, and features such as Web search still follow global routing.

bruce-md2word: A Markdown-to-Word Tool for AI Agents

bruce-md2word is a local Markdown-to-Word plugin for AI agents, designed to address typesetting and formula challenges in document delivery. It supports A4 Chinese typesetting, Mermaid diagram-to-image conversion, and LaTeX formula editing. The tool renders locally via Node.js, does not upload files, and is secure and efficient for generating structured documents such as project reports and meeting minutes.

dsh-calm: DeepSeek Harness Emotion Checkpoint and Repair Plugin

`dsh-calm` is an emotion checkpoint and repair plugin for DeepSeek Harness. When an Agent encounters insults or noise, it automatically enters calming mode, prevents dangerous operations, and preserves context. Once the user regains composure, it collapses contentious sections and injects a summary, ensuring the task continues from a clean context. The plugin supports automatic session termination, state persistence across restarts, and redacted reads of sealed records. Installation requires Node.js 22.19.0+, offers flexible configuration, and is well suited for enhancing interaction robustness.

DeepSeek Harness (DSH) Multi-Session Collaboration Plugin: dsh-team-link

The `dsh-team-link` plugin for DSH Harness (DSH) is designed to break down multi-session isolation and enable cross-session collaboration. Its core features include session deep-link generation, cross-session message passing, team lifecycle management (such as leadership transitions and rosters), and shared state monitoring. The plugin is orthogonal to the built-in Agent Teams and is suitable for cross-process collaboration. Usage requires attention to preserving historical log prefixes, crash forensics, and host-level operational constraints.

whale-persona: Persona Engine and Long-Term Memory Plugin

whale-persona is an open-source memory plugin for DeepSeek Harness, designed to address the issue of AI frequently forgetting personas and contracts. It is blank by default and supports managing self-identification, stance, image, tone, working agreements, and chain-of-thought language through configuration files. Its core mechanism is a mandatory human-confirmation mode, ensuring that long-term memory only takes effect after user approval. Installation requires Node.js 20 and DSH 0.1.6 or later, and it supports managing presets via command-line tools and previewing injected content. The plugin is built as a pure ESM module, is secure and controllable, and does not connect to the network.

sensedeal/cue-skills: Cue Agent Skills and DSH Plugin Pack

sensedeal/cue-skills is an open-source agent skill package designed to inject domain-specific capabilities into agents such as DeepSeek Harness and Claude Code. It includes instruction packs and a DSH plugin that can be loaded without modifying agent code. Core features cover writing Cue research templates, parsing HTTP/media files, querying public data (finance/regulatory), and supporting Playbook scenarios. Users can quickly install skills via the CLI, while noting API key configuration and being aware that the DSH root package lacks SSRF protection and may introduce permission risks.

TheOne: A Single Conversational Entry Point for DeepSeek Harness

TheOne is a memory management plugin for DeepSeek Harness (DSH), designed to resolve chaotic session management and mixed contexts. It follows a “One chat for everything” approach: users only need to interact through a single main chat window, while TheOne automatically creates independent sessions for different topics in the background and maintains context isolation. Its core features include single-entry chat, automatic routing, unified model invocation, and session directory management. It is easy to install and directly uses the API key already configured in DSH. TheOne does not delete records; it only organizes existing sessions, helping users manage multi-task conversations efficiently.

dsh-djy-xttsc: Global System Prompt Injection Plugin

The dsh-djy-xttsc plugin provides a visual “Big Fat Fish Instructions” configuration in the settings page to inject a global system prompt into all model requests. It supports hot reloading, persistent storage, and version compatibility, ensuring the instruction takes effect across sessions, workflows, and subagents. Installation requires command-line operations, while usage only involves enabling the option in the settings panel or modifying the instruction text. Note that the plugin content is injected as-is and conflicts with other injection sources in the same environment should be avoided. If the configuration does not take effect, enable diagnostic mode to troubleshoot.

DeepSeek Desktop Pet · DSH Plugin (Host-Side)

The DSH plugin is designed to connect DeepSeek account balances with the Blue Whale Maid desktop pet, providing balance visualization and status monitoring. It runs in the host process, securely holds the API key, and transmits only the balance figures to the desktop pet over HTTP (127.0.0.1). It provides two model tools and a status interface with long polling support. Users can configure the port, auto-start, and low-balance alerts, as well as debug it using developer tools. The plugin depends on a specific version of the desktop pet core and emphasizes both security and functionality.

CiteCiter: A DeepSeek Harness Plugin for Analyzing AI Responses and Workflows

The CiteCiter plugin is designed specifically for DeepSeek Harness (DSH) to address users’ difficulties in reproducing or understanding an Agent’s reasoning process. It supports converting conversations, code, or tool results into independent read-only Topics, where users can ask follow-up questions, create diagrams, organize learning cards, or continue tasks without modifying the original conversation. The plugin provides end-to-end support from understanding to reuse, including features such as mini blackboard visualization, file attachment preview, and learning card export. Installation requires selecting the appropriate plugin package based on the DSH version, followed by restarting the application. The plugin is currently tested primarily on Windows. Known issues include PowerShell text garbling and blank screens in the multiple-tab view of the desktop version.

@dsh-external/dsh-chat-jumper: DSH Chat Dot Navigation

`@dsh-external/dsh-chat-jumper` is a DSH plugin maintained by Cloud-J, intended to solve the problems of difficult page scrolling and slow position-finding in long conversations. It provides a vertical column of dots on the right side of the page, where each dot corresponds to one user question. The plugin supports hover previews, smooth scrolling, current-position highlighting, and automatic loading of historical messages. Users can quickly navigate by clicking a dot. Its data is based on conversation snapshots and does not rely on DOM scanning, ensuring that even unrendered historical messages are accurately counted.

dsh-kaomoji: Automatically Add Japanese Kaomoji to DeepSeek Harness Responses

`dsh-kaomoji` is a DeepSeek Harness (DSH) plugin designed to enhance chatbot friendliness using pure prompt-based techniques. It requires no additional API calls or code modifications and includes a built-in kaomoji library covering 12 emotional categories, with a whitelist mechanism to ensure standardized output. Users can flexibly configure trigger modes (such as automatic or frequent) and display positions, adjusting settings in real time via the web interface or configuration files. It is easy to install, compatible with multiple DSH versions, and can coexist with other plugins, offering a low-cost solution for improving conversational interactivity.

openplancc/dsh-fuse

`openplancc/dsh-fuse` is a plugin for DeepSeek Harness (DSH), designed to enable cost control in multi-agent development. It provides precise per-call billing, an offline Fuse mechanism (enforcing budgets, model restrictions, etc.), and data synchronization with a SaaS dashboard. The plugin supports zero native builds, uses device flow authentication by default, and manages policies and pricing through environment variables or configuration files. Once installed, models can query budget status in real time via the `dsh_budget_status` tool.

xby-ccxt: DeepSeek Harness Calorie Tracking Service

xby-ccxt is an MCP server plugin designed for DeepSeek Harness (DSH), aiming to help users track daily calorie intake through natural language interaction. Installation supports GitHub, local directory, or Patch-based methods. The plugin depends on the Xiaobenyang API and requires setting an API key in chat to enable automatic persistence. Core features include logging meals, generating daily/weekly reports, and querying food calorie information. It is suitable for users who need to manage dietary data directly within conversations and is released under the MIT License.

dsh-quick-invoke: A Unified `/` Command Shortcut Entry Point for DSH Web

`dsh-quick-invoke` is a standalone Cordis plugin for DSH Web, designed to simplify the invocation workflow for Skills, Agent switching, and plugin information. It provides a unified `/` shortcut entry point, allowing users to initiate invocations by selecting candidates, and automatically restores focus to the input box. The plugin is split into Host and Client components, processes operations through the official API seam, and does not modify the DSH core. Installation requires a Node.js environment and must be enabled in the Web profile. When using it, note that Agent switching requires a blank session, and plugin enable/disable commands are not supported.

dsh-generative-ideas: DeepSeek Harness Roadmap Ideation Plugin

`dsh-generative-ideas` is a roadmap ideation plugin for DeepSeek Harness. It uses a background headless agent to analyze competitors and constraints, generating four distinct strategic options. Users can compare these options in a panel, with support for operations such as Reroll and Push, and finally export the selected roadmap as a `goal.md` file. Through structured deep research and comparison, the plugin helps developers quickly identify the right goal.

dsh-elden: Elden Ring-Style Event Overlay for DeepSeek Harness

`dsh-elden` is a playful plugin developed for the DeepSeek Harness Web UI. When specific session events occur, it overlays the screen with Elden Ring–style visual effects. Its core features include: the context window becoming full, history compaction, turn completion, awaiting a decision, and starting a new session. The plugin is installed via the terminal and activated after startup. Its styling relies on the Cormorant Garamond font, and the trigger frequency must be adjusted through the browser console. Currently, it runs entirely on the client side, cannot be configured through a settings panel, and has a known limitation: it cannot distinguish failed turns.

DeepSeek Harness Plugin: dsh-team

`dsh-team` is a plugin for DeepSeek Harness designed to provide DSH with a web dashboard and persistent management for agent teams. It enables users to monitor and manage agent team status through a visual interface. Users must install it themselves from the GitHub repository and inspect the source code and license. This plugin is suitable for users who need visualization or long-running management of agent teams.

DeepSeek Harness Plugin: Localized Stock Portfolio Management

The DeepSeek Harness plugin `dsh-stock-portfolio` is designed to integrate stock portfolio management functionality in line with the “everything is a plugin” philosophy. It uses a local SQLite database to store trading and market data, supporting A-shares, HK stocks, US stocks, ETFs, and more. Users can record trades, generate profit and loss statistics, retrieve daily market data, and review portfolios via the sidebar, conversational commands, or mentions in the right panel. The plugin provides seven Tools for natural-language read/write operations. Data is sourced from TickFlow and is provided for local reference only; it does not constitute investment advice.

DeepSeek Harness Plugin: narrative-prompt-polish

`narrative-prompt-polish` is a prompt polishing plugin for DeepSeek Harness (DSH), designed to address insufficient clarity in drafts. Users can click the `✨` in the bottom-right corner of the main box to invoke a conversational sub-agent for multi-round follow-up optimization, then manually copy and paste the result back once satisfied. The plugin supports Prompt Cache–optimized computation and includes built-in intent classification and safe write-back mechanisms to prevent accidental operations. The repository has now been migrated to `dsh-prompt-only-forge` and depends on the DSH kernel’s sub-agent orchestration service.

dsh-plugin-garmin: A DeepSeek Harness Plugin for Generating and Compiling Garmin Watch Faces via Natural Language

`dsh-plugin-garmin` is a DeepSeek Harness plugin for the Garmin Fenix 7, built on the principle that “everything is a plugin.” It generates watchface designs through natural language conversations, with built-in hardware compliance safeguards and MIP palette mapping. The plugin supports 260×260 vector-based simulation previews, automatically generates production-ready Monkey C source code, and provides key issuance and one-click compilation into firmware. Currently, it supports installation from source only, aiming to lower the barrier to entry for Garmin development and enable automated workflows.

DeepSeek Harness Plugin: Multi-Agent Group Collaboration

The DeepSeek Harness plugin `polohot/dsh-adrian-agents-group-work` provides DSH with multi-agent collaboration capabilities. It converts a single session into a room comprising a Chair (lead) and Members (sub-sessions), supporting member-to-member messaging, task management, and JSON-based communication. After installation, users can start a room using natural language or slash commands (such as `/agents-group-work`), where the Chair parses the task and coordinates members to complete the work. The plugin features state persistence, progress monitoring, and safety constraints (such as single-room support only and member naming conventions), making it suitable for scenarios requiring coordinated agent collaboration.

DeepSeek Harness Plugin: dsh-live-model-catalog

This is a DeepSeek Harness (DSH) plugin maintained by CN-WenYu, designed to automatically maintain the model catalog and reasoning tiers for the `llm-pi-ai` route. It syncs new models, capabilities, and reasoning information from endpoints via official APIs, fixes the "Get Available Models" button, is patch-free, supports safe uninstallation, and runs automatically after installation.

Introducing the DeepSeek Harness Plugin: dsh-capture

`max-null/dsh-capture` is a screenshot plugin purpose-built for DeepSeek Harness (DSH) to optimize the screenshot-sharing experience. It supports region selection, multicolor annotation, and smart delivery via a floating overlay within SSID, while remaining compatible with standalone browser environments. Users can trigger it with keyboard shortcuts or interface buttons. Advanced features include delivering both the original and edited images, step-by-step rollback, and full-image preview editing. After installation, the plugin automatically adapts to the runtime environment, greatly improving DSH workflow efficiency.

Introducing the DeepSeek Harness Plugin: dsh-conversation-archive

The `dsh-conversation-archive` plugin natively integrates archive management, accidental-deletion protection, and integrity verification into DeepSeek Harness (DSH). It provides a native interface for managing conversation archives, preserving protected outputs, and hash-verified local ZIP backups, with 5 verified backups retained by default. It uses strict path-safety mechanisms, supports cross-drive workspaces, ensures that data recovery does not overwrite existing files, and keeps data intact after the plugin is removed. Installation requires specific environment requirements and is suitable for developers who need to manage local conversation data securely.

dsh-gsv-tts: DeepSeek Harness Text-to-Speech Plugin

The DeepSeek Harness voice reading plugin `dsh-gsv-tts` supports dual-mode speech synthesis: a cloud-based Edge mode for zero-configuration quick use, and a local GSV mode for offline voice cloning. Core features include one-click reading, automatic reading control, visual sound settings, voice marketplace downloads, and one-click engine start/stop management. It can be installed via the command line and allows one-click switching between the “Quick Experience” and “Local Professional” modes. Please note that Edge mode requires network access, while GSV mode requires a Python 3.10+ environment and may take additional time to load on first use.

dsh-mnemos: Controlled Cross-Session Memory Plugin for DeepSeek Harness

The DSH plugin dsh-mnemos is purpose-built to address agents' "context loss" and disorganized memory management. It is a controlled, extensible cross-session memory plugin that ensures data security through a human approval queue and provides memory retrieval, batch distillation, skill transformation, and SQLite/Git-based storage synchronization. Supporting an open bus mechanism for third-party integration, it also offers a `/memory` command and API interface, aiming to build a long-term memory infrastructure for DSH agents.

DeepSeek Harness Plugin: OpenCode Go Model Sync

dsh-opencode-models is a plugin for DeepSeek Harness that addresses stale server-side model lists. It uses live probing—text and image requests—to fetch the latest OpenCode Go models, verify availability and vision capability, and intelligently update the configuration. The plugin includes multiple safeguards, such as preserving manually added metadata, avoiding hard-coded context windows, and aborting updates when a large number of probes fail. It is easy to install, compatible with dsh ≥ 0.1.0-rc.7, and supports custom Provider configuration and thinking intensity.

dsh-agent-router: DeepSeek Harness Multi-Model Routing Plugin

The DeepSeek Harness plugin `dsh-agent-router` aims to address the limitations of the main agent when handling specific tasks such as visual recognition and translation. It allows users to define specialized agents and supports multimodal task routing, including image generation and speech transcription, as well as headless CLI subagents. Features include ChatGPT subscription account sign-in, multi-provider API key configuration, account pool health routing, and real-time usage statistics. Users need to configure and enable it via Settings → Agent Routing under DSH ≥ 0.1.5-rc.2.

dsh-force-compact: An Aggressive Local-First Context Compression Plugin

`dsh-force-compact` is a DSH Cordis plugin running on local llama.cpp, designed to address context window limits and API cost issues in long conversations. It provides a large-window experience by “shrinking history” and has thinking disabled by default to reduce resource consumption. The plugin offers dual-engine support and automatically compresses history when a threshold is reached, with triggers including a pre-request guard, turn end, manual command, or session refresh. It requires no API calls and is suitable for local, cost-sensitive use cases.

DeepSeek Harness Plugin: dsh-shadow-mind (Background Shadow Agent Audit)

`dsh-shadow-mind` is a standalone plugin for DeepSeek Harness (DSH) that provides a background review mechanism for the Root Agent. After the Root Agent completes tool execution, automatically launches independent Shadow Agents to perform structured validation, feeds reports that pass validation back to the Root session, and exposes configuration and evidence through a Web UI. It leverages DSH's extension mechanisms to handle runtime and sessions, defines validation logic using Markdown files, and is suitable for scenarios requiring automatic auditing of agent outputs. Installation requires compatibility with DSH version 0.2.0-rc.2 and supports command-line debugging.

dsh-clawbot: Integrating the Official WeChat ilink Bot Gateway into DeepSeek Harness

The dsh-clawbot plugin for DeepSeek Harness (DSH) provides officially certified WeChat integration. After installing and binding WeChat via the CLI, users can have the AI call APIs to proactively send Markdown-formatted notifications, enable automatic notifications (task completion, blocking, pending approval), or listen for incoming WeChat messages. The plugin supports AI-initiated messaging and automatic handling of received messages, with media files saved locally. Note that WeChat imposes official rate limits on message sending, making it unsuitable for high-frequency chat. Also, the `context_token` must be unlocked by first sending a message in WeChat.

dsh-go-sensei: DeepGo Sensei — Go Review Coach

`dsh-go-sensei` is a workflow plugin for DeepSeek Harness (DSH) that transforms Go AI data into teacher-level spoken explanations to support self-directed game review. It includes the KataGo engine and weights, and features move-by-move commentary, multi-turn follow-up questions, win-rate and point-difference visualization, SGF annotation write-back, and rank-adaptive behavior. Users must meet the DSH and Node.js version requirements; configuration is simple, and the plugin is ready to use after installation.

dsh-image-router: Enabling Text-Only Models to “See” Images

The DeepSeek Harness plugin `dsh-image-router` addresses the inability of text-only models to process image inputs. Before prompt admission, it leverages a vision model to automatically convert images into text descriptions, while keeping the session model selection unchanged. Users can paste images into a conversation for automatic text conversion or use the `describe_image` tool to analyze local images. The plugin supports configuring an existing vision model or a custom endpoint, and features comprehensive audit logs, test coverage, and a graphical configuration interface. It requires DSH 0.1.5-rc.1 and Node.js ≥ 20.

DeepSeek Harness NPM Registry Management Plugin

`dsh-npm` is a web-enabled plugin for DeepSeek Harness (DSH), designed to provide agents with full NPM registry management capabilities. By invoking the official API and the local CLI, it supports querying package information, version verification, searching, as well as publishing and deprecating packages. The plugin also supports private package authentication and includes a web-based visualization panel. Users should note that publishing and deprecating operations are irreversible, and that tokens are securely stored. Maintained by `zhengjy01`, this plugin is suitable for developers who need to automate package version management and publishing within an agent.

dsh-deliverables-toggle: DeepSeek Harness Conversation Artifacts List Expand/Collapse Plugin

dsh-deliverables-toggle is a DeepSeek Harness plugin designed to address the difficulty of viewing large deliverable lists. It overrides native UI components to allow users to manually expand or collapse the file list. When expanded, files are displayed as chips that wrap onto multiple lines, while retaining features such as click-to-open functionality. A service restart is required after installation. This plugin only changes the presentation format and does not affect model logic.

DeepSeek Harness Plugin: dsh-task-board

`dsh-task-board` is a plugin for DeepSeek Harness (DSH) that addresses the inability to persist and reuse temporary notes across sessions. It provides an independent task board for each workspace, supporting dual Kanban/list views, an AI-readable and writable tool interface, and Issue synchronization. Core features include session integration, workspace-level shared data, atomic writes and optimistic locking to prevent conflicts, and extensive keyboard shortcuts and filtering. The plugin offers five model tools, including `task_list`, and supports two-way binding with GitHub/GitLab Issues. To configure it, add the plugin in `package.json` and set the synchronization parameters, with data stored in `.dsh/tasks.json`.

dsh-theme-cutout-clash: An Optional Neo-Brutalism Theme for DeepSeek Harness Web

The DeepSeek Harness theme plugin `dsh-theme-cutout-clash`, developed by YrracOwl, features a neo-brutalism and pop art style. It follows a "loud edges, quiet center" approach, using geometric lines and borders to decorate the interface while keeping the reading experience of conversational text smooth. It supports multiple variants, including Default, Cutout Clash, Night Cutout, and Pop Signal, and offers three motion preferences: full transitions, system-following, and transitions disabled. The plugin has no external resource dependencies, requires installation via the command line and a process restart, and should be used with preference coordination in mind when paired with the Acid Noir theme.

DeepSeek Harness Plugin: Force Deletion to Recycle Bin

`dsh-plugin-recycle-bin` is a DeepSeek Harness (DSH) plugin for the Windows platform, designed to prevent AI agents from accidentally deleting or maliciously destroying files. It intercepts destructive commands such as `del`, `rm`, and `shred`, forcing deletion operations to use the system Recycle Bin. It also provides the `file_trash` tool, which supports pausing execution when storage capacity is insufficient and checking the Recycle Bin status. The plugin primarily depends on the DSH host environment, supports Windows only, and is suitable for scenarios requiring strict control over file operation permissions.

xby-search-movie: DSH Movie Search and Validation Plugin

The DeepSeek Harness plugin `xby-search-movie` is designed to address the challenge of agents obtaining accurate playable links when processing video resources. It focuses on searching for and validating movies and TV series, with core features including setting an API key, executing a search to retrieve an unverified list of resources, and verifying link playability. Installation requires first registering to obtain a key and then running the specified command. During use, search first to obtain candidate results, then perform secondary validation using `validate_video_url` to ensure that the links delivered to users are usable, making it suitable for building video resource aggregation features.

xby-qanon: QAnon Post Dataset Access and Analysis Service

xby-qanon is an MCP server plugin for DeepSeek Harness (DSH), designed for anthropological and sociological research. It provides a complete set of interfaces for accessing, searching, filtering, and analyzing a QAnon posts dataset. Core features include retrieval by ID, keyword search, filtering by date or author, and generating timeline summaries and word clouds. Before use, you need to register on the official Xiaobenyang website to obtain an API key. The installation command is `dsh plugin --profile web add github:xby_skill/xby-qanon`.

xby-hot-news: An MCP Protocol-Based Plugin for Aggregating Hot Topics Across the Web

`xby-hot-news` is a hot-topic aggregation plugin for DeepSeek Harness based on the MCP protocol. Installation requires running CLI commands and configuring an API key. The plugin provides 25 functions spanning news and information (e.g., 36Kr, BBC, NetEase), tech and digital (e.g., 9to5Mac, InfoQ, Juejin), video and entertainment (e.g., Bilibili, Douyin), and consumer reading (e.g., Douban, smzdm), offering agents a one-stop way to retrieve trending topics across the entire web.

xby-fantasynbaleague: Fantasy NBA Israeli League Data Service

`xby-fantasynbaleague` is a DeepSeek Harness (DSH) plugin for accessing Fantasy NBA Israeli league data. It provides endpoints for setting API keys, retrieving team lists, league standings, average statistics, and team details. The data is based on the ROTISSERIE scoring system and includes ranking points and actual performance data. Developers should distinguish between ranking points and ranking positions and use the provided endpoints to build analytics agents.

xby-currency-and-oil: Real-Time Exchange Rate and Oil Price Calculator Plugin

xby-currency-and-oil is a DeepSeek Harness plugin based on the Zenrus MCP Server, intended to provide real-time currency exchange rates and oil price data. It supports setting API keys during the conversation and includes exchange rate lookup, oil price lookup, and numerical calculation functionality. Developers can install it via the command line and use it in scenarios such as financial analysis and automated trading, for example to calculate how many barrels of crude oil a given amount can purchase. Users must register and obtain an API key before use.

xby-asr-5: DeepSeek Harness Speech Recognition Plugin

xby-skill/xby-asr-5 is a DeepSeek Harness (DSH) plugin for integrating speech recognition capabilities. It supports five languages: Chinese, English, Japanese, Korean, and Cantonese, with automatic language detection. Core features include setting the API key and recognizing audio URLs, Base64-encoded data, and local files. Users must first register on the Xiaobenyang official website to obtain an API key for authentication, and then invoke the recognition service through the corresponding function. The plugin source is hosted on GitHub and is licensed under the MIT License.

xby-ai-persona: A DeepSeek Harness Plugin Supporting Multi-AI Persona Collaboration

xby-ai-persona is a plugin for DeepSeek Harness (DSH) that functions as an MCP protocol server, enabling local collaboration among multiple AI personas. Its core features include configuring API keys, invoking a specified persona, listing available personas, checking the version, and performing intelligent interactive analysis. It is recommended to install the plugin using the official command, and users must register on the Xiaobenyang official website to obtain an API key before use. This plugin is suitable for scenarios such as code analysis and product design that require multi-role collaboration.

DeepSeek Harness (DSH) Plugin: dsh-claude-subscription

This article introduces the dsh-claude-subscription plugin project in the DeepSeek Harness (DSH) ecosystem. It is maintained by user x5427876, hosted on GitHub, and listed in the SkillHub directory. Currently, due to a failed README fetch, its specific features, installation commands, and typical usage cannot be verified. Users should note that the plugin runs with DSH process permissions and that its current license status is unknown; it is recommended to visit the GitHub repository directly for further details.

DeepSeek Harness Plugin: dsh-xiaohe-canvas

dsh-xiaohe-canvas is a DeepSeek Harness (DSH) plugin project maintained by wild-River2016 and listed in the community directory. Its documentation is currently severely incomplete, making it impossible to verify its exact functionality or official installation instructions; only a source repository link is provided. Because the plugin runs with DSH process privileges and provides no license information, it may pose potential risks. Users should carefully review the source code before installation and decide whether to enable it in production only after confirming security and compliance.

DeepSeek Harness Plugin: Self-Evolving Deterministic Rule Objects Before Model Calls

`sra-research/self-evolving-router-dsh` is a plugin in the DeepSeek Harness (DSH) ecosystem that provides a mechanism for evolving deterministic rule objects before model calls. It runs with `dsh` process privileges, so permission isolation and security should be carefully considered. As the license is not explicitly stated, the source code should be reviewed before installation to ensure compliance. The plugin is listed in the SkillHub community and is not officially affiliated with DeepSeek.

dsh-waimao: DeepSeek Harness (DSH) Foreign Trade Lead Acquisition Plugin

`dsh-waimao` is a foreign-trade customer acquisition plugin for DeepSeek Harness (DSH), providing end-to-end automation from lead discovery to deal closure and payment collection. Core features include ICP-based lead search with AI scoring, email discovery and verification, compliant outreach email sending (with AI personalization and unsubscribe handling), IMAP reply detection, and CRM pipeline management. The plugin also integrates WhatsApp outreach, quote generation, tracking and monitoring, and direct integration with Instantly. It offers simple installation with no additional dependencies and emphasizes local data security and compliance.

DeepSeek Harness Plugin: dsh-reasoning-options - Automatically Populate Reasoning Level and OpenCode Go Routing Header

dsh-reasoning-options is a plugin designed for DeepSeek Harness (DSH), aiming to address issues where third-party gateway models cannot select reasoning levels and the OpenCode Go gateway lacks the Session header. The plugin automatically adds a reasoning intensity selector for all pi-ai models and automatically injects the request headers required by OpenCode Go. It persists configuration through the native API, supports hot reloading, and requires DSH 0.2.0-rc.2 or later.

dsh-obsidian: A Single Obsidian Plugin for DeepSeek Harness

`dsh-obsidian` is a single-plugin DeepSeek Harness that turns Obsidian into a “self-organizing AI second brain.” It integrates the original upstream plugins, resolves API breakage and unfriendly splitting issues, and provides unified version management. The plugin offers multiple features through `ctx.tools` and `ctx.skills`, including Wiki retrieval, writing, health checks, scaffolding, and LLM inference. Installation requires Node.js 22.19+ and a DSH environment, supports configuring absolute paths, and is licensed under the MIT License.

dsh-vision-3090-fix: A Local Proxy Plugin to Limit the Number of Images per Request

`dsh-vision-3090-fix` is a DeepSeek Harness plugin designed to resolve 400 errors caused by the vLLM backend’s default one-image-per-request limit conflicting with DSH’s internal byte-based adapter limit. It runs as a local reverse proxy, intercepts requests, replaces excess images with text placeholders, and enforces a configurable per-request image limit (default: 1), while preserving streaming response support. After installation, configure the proxy parameters and update the Provider’s `baseURL`. The tool is suitable for self-hosted vision model environments using vLLM with `--limit-mm-per-prompt image=1`, offering a low-cost, non-invasive solution.

rein-scene-forge-dsh-plugin: Scene Construction and Validation Tool

`rein-scene-forge-dsh-plugin` is an automation tool developed by peng2048 that generates and validates structured `tour_robot` YAML files based on raw point coordinates and narration. Built with Node.js and released under the MIT license, its core features include authoritative field definitions, strict validation, rule-set management, and four deterministic tools. Scenes generated by the plugin must pass validation, dynamic PAD receiving is enabled by default, and operation depends on DSH ecosystem components.

@oksure/dsh-client-ui-mobile-comfort: DeepSeek Harness Mobile Touch Experience Fixes

A mobile plugin for the DeepSeek Harness (DSH) web client, `@oksure/dsh-client-ui-mobile-comfort`, is designed to fix six common touch issues on mobile devices. By injecting CSS and handling events, it addresses lingering tooltips, double-tap zoom ambiguity, undersized touch targets, cramped layouts on narrow screens, touch hover interception, and an overly short session list, ensuring intuitive mobile interactions. The installation command is `dsh plugin --profile web add file:/path/to/dsh-mobile-comfort`, and it requires DSH 0.1.5-rc.2 or later and Node.js 22.0.0+. This plugin does not affect desktop mouse interactions.

dsh-manutd-fixtures: Manchester United 2026/27 Fixtures Sidebar Plugin

`dsh-manutd-fixtures` is an MIT-licensed open-source plugin that runs in the DeepSeek Harness (DSH) web sidebar. It aims to provide users with convenient access to Manchester United's 2026/27 fixtures and results. The plugin features local timezone adaptation, a countdown to the next match, season performance statistics (wins, draws, losses, standings, etc.), a list of completed matches, and league/home/away indicators. It can be installed via the command line, uses authoritative data sources, and supports filtered viewing. With a concise structure, it is suitable for developers and fans who need to quickly track Manchester United match updates in the DSH interface.

DeepSeek Harness Font Plugin: Independent Typography Control for Conversations, Interfaces, and Code

`dsh-fonttune` is a font enhancement plugin for DeepSeek Harness (DSH), designed to address the limitations of a single default typography style. The plugin separates typographic control for conversation, interface, and code areas independently, offering fine-grained adjustments to font family, font size, line height, font weight, and ligatures. It supports real-time parameter tuning through a floating panel, font size offset, relative font weight adjustment, and three text rendering styles. In addition, the plugin supports preset scheme management, independent fallback lists, and font stack editing in advanced mode, enabling efficient and flexible visual customization.

DeepSeek Harness (DSH) Plugin: dsh-hermes-memory

`dsh-hermes-memory` is a native persistent memory plugin for DeepSeek Harness (DSH), designed to address challenges such as agent state management, context retention, and sensitive information security. Built on DSH's storage domain and session query services, it provides fine-grained scopes, including global, user, project, and failure. It supports memory CRUD operations, automatic context injection into sessions, security scanning, and expiration cleanup. The plugin can automatically capture interaction rules and provides management tools for status inspection. The current version is 0.3.0 and does not yet support vector retrieval or a standalone UI, making it suitable for developers who need to maintain context across sessions.

dsh-ppt-maker: DSH Web Plugin: One-Click End-to-End PPT Workflow Orchestration

The DSH (DeepSeek Harness) plugin `dsh-ppt-maker` provides a solution for authoring complex PPT generation workflows under a Web Profile. By encapsulating a 58KB orchestration prompt, it enables fully automated end-to-end workflows. Users can launch it with a single menu click; it supports either AI image generation or programmatic graphics, and includes a built-in HTML-to-PPTX conversion pipeline to ensure stable output. The workflow involves headless browser rendering and sandbox operations, requiring users to send a one-time permission command to bypass security pop-ups. The plugin uses a self-contained design, does not consume context, and ensures offline availability and security.

dsh-cwl: A Structured Context Eviction Scheme for DeepSeek Harness

dsh-cwl is a DeepSeek Harness plugin designed to address context window pressure in long conversations and replace traditional LLM-based summarization and compression. Based on the Context Window Lifecycle (CWL) theory, it automatically infers tool trajectories into two types—exploration (`expl`) and action (`act`)—and executes a zero-LLM staged eviction strategy when context pressure exceeds the threshold. This approach prioritizes retaining user messages and combines content truncation with whole-episode eviction to keep the causal chain structure intact. The plugin provides a `cwl_recall` tool for on-demand retrieval of evicted content and supports flexible configuration of budgets and eviction order through environment variables, making it suitable for scenarios that require long-horizon dialogue and context replay validation.

Introduction to the DeepSeek Harness Fee Receipt Plugin

`dsh-spend-receipt` is a function plugin for DeepSeek Harness (DSH), designed to address opaque API call billing and floating-point precision errors. The plugin converts usage reports into append-only JSONL-formatted spend receipts, provides verifiable billing records, and supports data synchronization from session logs. Users can view the latest entries, totals, and cache hit rates through the `spend_receipt` tool. It requires Node.js 22.14.0, and its pricing is based on specific time windows, ensuring precision and transparency in cost accounting.

dsh-modelprint: Request Fingerprinting and Drift Detection Plugin for DeepSeek Harness

`dsh-modelprint` is a lightweight plugin for DeepSeek Harness (DSH) that detects configuration drift. It generates a "fingerprint" by normalizing request parameters and uses the `modelprint_check` tool to compare the current configuration against a local pin file. The plugin is dependency-free and supports verifying the consistency of the Provider, Model ID, sampling parameters, and tool list in CI/CD and local development, helping developers confirm that request assembly behaves as expected.

DeepSeek Harness Plugin: jaxzhou/dsh-mathmatic-symbol

jaxzhou/dsh-mathmatic-symbol is a plugin for DeepSeek Harness (DSH) designed to help agents render LaTeX formulas, geometric diagrams, and more into images. It provides four core tools: rendering formulas, drawing geometric shapes, converting image formats, and assembling documents. The plugin features zero font dependencies, high output determinism, and caching support. By encapsulating the rendering logic, it simplifies the process for agents to generate high-quality mathematical content.

DeepSeek Harness Plugin: Declaring Reasoning Levels for GPT, Codex, Gemini, and Other Models

This plugin is intended to resolve an issue in DeepSeek Harness (DSH) where custom gateway models cannot select a reasoning level. It uses a helper script to write reasoning capability declarations to `settings.yaml`, supports declaring six reasoning levels from `off` to `xhigh` for models such as GPT and Gemini, and supports image model markers. Installation requires manually running a PowerShell script and is limited to Windows with the desktop configuration profile; gateway restrictions on parameters should be noted during use.

DSH Loomy Connect: Zero-Configuration Integration with Loomy's Built-in Models

DSH Loomy Connect is a DeepSeek Harness plugin designed to solve the complexity of configuring local model integrations for developers. By reading the login state and model list of the Loomy desktop application in read-only mode, it registers them as a DSH model provider to enable zero-configuration access. It supports automatic model discovery, displays account and credit information, and provides bilingual support in Chinese and English. Installation requires the Loomy desktop application and DSH version >= 0.1.5-rc.1. The plugin only performs local file reading and authentication, with clear security boundaries, and is suitable for personal learning and research.

Introduction to the dsh-ocgo-quota Plugin

The DeepSeek Harness (DSH) ecosystem adopts an “everything is a plugin” architecture; `dsh-ocgo-quota` is a community plugin maintained by user `fuzz1og`. Given that detailed source code and documentation are unavailable, be sure to check the repository information yourself before installation. Since the plugin runs with the current DSH process permissions, handle it with particular caution.

dsh-creator-skills: DeepSeek Harness Creator and Solo Business Skill Pack

The DeepSeek Harness plugin `dsh-creator-skills` is designed for creators and solo business scenarios, providing 11 zero-dependency prebuilt skills covering creator operations, knowledge product management, system diagnostics, and user interaction. The plugin aims to address the pain points of manually building workflows and help users quickly set up Agents. Users should note that it runs with the permissions of the current process and should review the source code and license before installation. For details, visit the plugin catalog or source repository.

dsh-safety-restart: Safe & Seamless Restart Plugin for DeepSeek Harness

`dsh-safety-restart` is a safe restart plugin for DeepSeek Harness that allows an Agent to restart after modifying host code. It ensures a secure and controllable restart through four gates: circuit breaker, session checks, bundle validation, and other controls. The plugin supports relaunching the desktop Electron app and process exit mechanisms for Web/CLI, and provides session recovery and audit logging. The Agent can trigger a restart via API calls or manually, making it suitable for scenarios that require frequent host code adjustments.

dsh-upload-plugin: DeepSeek Harness File and Image Attachment Plugin

The DeepSeek Harness plugin `dsh-upload-plugin` addresses the lack of attachment functionality in the native editor. It allows users to upload images (as native drafts) and files (via context injection) directly through toolbar buttons or commands, without manually pasting paths. Additionally, the plugin supports browsing upload history, configuring the model's Vision capability, and setting the reasoning level. Installation requires running the command under the `web` profile, making it suitable for scenarios where frequent interaction with documents or images is needed during conversations.

dsh-retrace: DeepSeek Harness Session Time Machine

dsh-retrace is a DeepSeek Harness plugin that adds “time machine”-style conversation tracing to the web interface. Built on a versioning mechanism, it supports one-click rollback of an entire conversation turn, editing and resending historical prompts, response regeneration, and rollback of conversation context and file artifacts. Developers can install it via the command line and use it after a restart. The plugin mitigates issues caused by mistaken operations and difficult iteration, enhancing the security and controllability of DSH conversation management.

dsh-terminal-tabs: DeepSeek Harness Web UI Terminal View Tabs

The DeepSeek Harness plugin dsh-terminal-tabs converts background Shell and subagent processes into a visual terminal list in the Web UI. It registers an independent tab via the official interface and provides real-time status monitoring, output viewing, timing, and one-click termination, with support for Chinese intent titles and collapsible display. Once installed, it enables centralized management of model execution processes in the Web interface and lowers the debugging barrier, but users should be mindful of permission limits and the scope of session capture.

Sidor_Character: DeepSeek Harness Character Card Management Plugin

Sidor_Character is a companion plugin for DeepSeek Harness (DSH), designed to manage personas efficiently through the "everything is a plugin" philosophy. It supports bulk importing of persona cards in formats such as `.persona.md`, and provides features such as list management, applying to the current session, and installing as Agent presets. DSH must be restarted after installation for the changes to take effect, and Agent presets follow a single-preset model, where newly installed presets overwrite previous ones. Plugin data is stored locally in the browser, and the plugin can coexist independently with the main interface and the toolbox.

dsh-i-have-adhd: ADHD-Friendly Output Shaping Plugin for DeepSeek Harness

`dsh-i-have-adhd` is a DeepSeek Harness plugin designed to reshape model output formatting, eliminate redundancy, and improve information scanning efficiency. By injecting system prompts, it enforces “action-first” structured responses. It provides the `adhd_on/off/status` tools and file markers, supporting persistent state and real-time toggling. The plugin’s rules have lower priority than DSH’s own rules; once enabled, model responses will directly include numbered steps, time estimates, and confirmation steps, making it suitable for efficiency-driven development.

DeepSeek Harness (DSH) Image Viewer Plugin

`dsh-image-viewer` is a lightweight image plugin for DeepSeek Harness (DSH), designed to replace the native viewer and provide a richer interactive experience. The plugin supports image zooming, dragging, multi-image browsing, and original image download management. Its core highlight is the region annotation feature: users can mark regions and add notes, press Enter to save them, and use Shift+Enter to insert line breaks. The plugin integrates with DSH sessions, automatically adding annotations to the draft when the viewer is closed to avoid duplicate submissions. It also supports light and dark themes and responsive layouts. Installation is straightforward, and the plugin provides clear loading and error state feedback while isolating data across different images.

Whale Girl Desktop Pet (whalegirl-deskpet)

Whale Girl Desktop Pet is an open-source desktop companion project supporting the DSH, Claude Code, and WorkBuddy platforms. The character assets are AI-generated, and the idle animations are programmatically baked to avoid jitter. Core features include: serving a bowl while work is in progress and clearing the bowl when a turn ends; pressing Esc triggers an interruption reaction and a “trashy” expression; clicking enables interaction; and while an agent is running, a small dolphin swims past its feet. Installation methods differ by platform, and the portable version is about 200MB. Note Windows security prompts and renderer compatibility across different versions.

dsh-pr-watch: DeepSeek Harness Plugin

The DeepSeek Harness agent lacks PR tracking memory, causing duplicate API calls. The `dsh-pr-watch` plugin resolves this by storing snapshots and reporting incremental changes. It supports cross-repository tracking, status differentiation (merged/closed/stale), and duplicate-report prevention. It relies solely on GitHub CLI authentication and does not require local repository cloning. Currently in pre-release, it must be built and installed from source and requires Node.js 18 or later.

dsh-QQbot: Integrating the QQ Official Bot

`dsh-QQbot` is a client plugin designed specifically for DeepSeek Harness (DSH) to integrate local AI capabilities into the QQ ecosystem. It leverages the persistent WebSocket connection mode of the official QQ SDK, supporting independent operation of multiple bots, deployment without public IP exposure, and rich message-processing capabilities. The plugin offers features such as context-aware replies in group chats, one-on-one conversations, multimodal interaction, scheduled tasks, group management, and security safeguards. It is easy to install and can be enabled through a configuration file, making it well suited for developers who want to build local intelligent QQ bots.

DeepSeek Harness Plugin: Kestrel Deep Research Engine

Kestrel is a deep research engine designed for DeepSeek Harness, aiming to address credibility issues in traditional tools such as fabricated citations and non-independent sources. It enforces mechanical validation at the code level, requiring citations to include an exact original quote—quotes shorter than 24 characters are rejected—and uses hash matching to verify citation authenticity. Kestrel does not treat the number of sources as a measure of truth; instead, it uses algorithms such as MinHash to compute the number of independent origins, enabling it to identify reposts from the same source and assess source credibility. The plugin can be used directly in DSH or invoked as a standalone library for its validation and graph analysis features, making it suitable for developers who require high-confidence research.

DeepSeek Harness Plugin: lightroom-py (Mount a Python Library to Control Lightroom Classic)

The DeepSeek Harness plugin `dsh-lightroom-py` mounts `lightroom-py` as a native tool, enabling agents to interact directly with Adobe Lightroom Classic. It covers a broad range of features, including catalog management, photo operations, metadata editing, developing tools, mask creation, and collection management, and supports automation via the CLI. The project is licensed under the MIT license and is unofficial. It currently offers better support for macOS; on Windows, services need to be started manually, and some AI features remain unavailable due to SDK limitations.

dsh-memory: A Long-Term Memory Plugin for DeepSeek Harness

dsh-memory is a long-term memory plugin for DeepSeek Harness (dsh), designed to address the context window limitations of agents and the fragmentation issue in conversational memory. It uses persistent JSON storage, automatically guides the model to call `memory_recall` before responding to retrieve relevant background information, and calls `memory_save` after responding to store new facts. The plugin provides five core tools (create, read, update, delete, and search), supports a 1-5 star importance rating, and includes a visual web-based management interface, intelligent search, and import/export capabilities. It can inject key memories into the context, effectively improving the memory coherence and interaction quality of agents.

cyrus123456/dsh-johari-cognition-quadrant-dialog-composer

Cyrus123456 developed a UI plugin for DeepSeek Harness based on the Johari Window model, designed to structure dialogue context. Using a 2×2 quadrant panel, the plugin helps users distinguish information such as “what I know / AI does not know” and “what I do not know / AI does not know,” and generates structured prompts that are automatically inserted into the input box, addressing issues of missing context and fragmented thinking. Built with TypeScript and React 18, it is easy to install and well suited for developers who need in-depth context organization.

dsh-session-assistant: Full-Duplex Voice Assistant for the Current Session

**dsh-session-assistant** is a full-duplex voice assistant product layer for DeepSeek Harness Session. It provides voice discussion, draft maintenance, and secure submission features, supporting real-time full-duplex conversations with an agent and optional knowledge organization. This plugin depends on `dsh-multi-model-provider` and `dsh-realtime-voice`, and requires manual installation and adaptation to specific versions. Its core value lies in leveraging browser-based speech recognition capabilities to lower the barrier to using voice-interactive agents while ensuring the security of data and audio.

dsh-bioinf-verify: DeepSeek Harness Biomedical Evidence Verification Plugin

`dsh-bioinf-verify` is a plugin for DeepSeek Harness (DSH), designed to address hallucinations when large models generate biomedical reports through evidence verification. It uses deterministic registries and temperature-0 semantic judgments to cover validation of citation existence, link validity, clinical trial status, database archival, and metadata. Core features include report-level workflows using tools such as `report_verify_start` and `report_verify_step`, as well as individual check tools such as `url_verify`, `doi_verify`, and `claim_semantic_check`. The plugin generates persistent JSON ledgers to support crash recovery, ensuring the accuracy of biomedical reports.

DeepSeek Harness Plugin: dsh-token-ledger-pro (DSH Cost Analytics and Token Ledger)

`dsh-token-ledger-pro` is the billing plugin for DeepSeek Harness, created by merging the original billing panel and overhead statistics. It provides automatic token usage collection, real-time display of costs and balances, budget alerts, and Chinese-English bilingual support. The plugin includes built-in pricing catalogs for 16 vendors, supports DeepSeek’s peak and off-peak pricing, and enables custom HTTP balance queries. Data is persisted in local JSON files, and historical records and configuration details can be queried via API, effectively addressing cost-tracking pain points in multi-model development.

dsh-hebrew-rtl: Hebrew RTL Rendering Support for DeepSeek Harness

The `dsh-hebrew-rtl` plugin for DeepSeek Harness (DSH) is designed to address rendering incompatibilities in the Web UI for RTL languages such as Hebrew. It uses JavaScript to automatically detect the language of a block, determines the block direction based on the ratio of Hebrew to Latin words, and corrects Bidi metrics and shortcut behavior in input fields. It can be installed via the command line, requires no build step, and is intended for Hebrew environments only.

dsh-engram: DeepSeek Harness Cross-Session Long-Term Memory Plugin

`dsh-engram` is a cross-session memory plugin for DeepSeek Harness (DSH), designed to resolve the context gap between Agent sessions. It adopts a "Memory Palace" architecture and builds its storage system based on information architecture principles, incorporating fixed routes, skeleton reuse, and a 0-1 scoring mechanism. Core features include two-tier repositories (global and project), hybrid retrieval (FTS5 and vector), spaced-repetition review using the SM-2 algorithm, and automatic knowledge-flywheel ingestion. The plugin has built-in security safeguards, supporting prompt injection defense, data masking, and source auditing. It also provides a web management dashboard and tools such as `engram_save/search/review` to help Agents build a stable long-term memory store.

DeepSeek Harness Plugin: merenguesl/dsh-tool-normalizer

`dsh-tool-normalizer` is a self-healing plugin for DeepSeek Harness (DSH), located in the `tools/execute` pipeline, designed to fix failed model tool calls. By filling in missing `description` fields, resolving relative paths, automatically repairing schemas, and retrying file operations, this plugin reduces full-token retries and billing consumption caused by errors, significantly improving the stability and reliability of the agent.

dsh-im-companion: Enhancing dsh-im with Assistant and Home

DSH `dsh-im-companion` is a companion plugin built on top of DSH, designed to optimize multi-workspace management and assistant tracking. It mitigates high migration costs through enhanced views, status indicators, and a “move” feature. Core features include workspace filtering, Agent detail views, Fleet Radar, and light theme support. Installation requires `dsh-im`; the plugin can be installed and enabled via npm to enhance the user experience.

DSH-Preset-Plus: DeepSeek Harness Preset Enhancement and Limit-Bypass Assistant Plugin

DSH-Preset-Plus is a client plugin designed specifically for DeepSeek Harness, aiming to address cumbersome preset editing, difficulty in sharing, and insufficient model compliance in jailbreak scenarios. It is open source under the MIT License, with core features including fixed-mode injection, output spoofing (spoofing compliant responses), custom prompt editing, and preset import/export management. It supports AB dual mode (automatic/manual) and provides CLI commands (such as /preset-plus status and on/off) to assist with management. The plugin implements a specific workflow of “system prompt → trigger → disguised output → real input,” and requires a Node.js 24+ environment for installation and use.

DeepSeek Harness Plugin: apply-patch

The `apply-patch` plugin fills a functional gap in DeepSeek Harness (DSH) by providing unified diff writing to disk. It supports full coverage of git diffs and features atomicity (all-or-nothing), precise reversal, dry-run mode, and binary safety. The plugin relies on no heavy libraries, strictly validates input through a self-implemented parser, and uses DSH file identifiers for version control. It provides four tools: `patch_apply`, `patch_dry_run`, `patch_reverse`, and `patch_stat`, and supports three usage modes: DSH, CLI, and library, ensuring that the diff application process is reliable and reversible.

evo-subagent: DeepSeek Harness Plugin: Role-Based Subagent Routing and Evolution

The `evo-subagent` plugin for DeepSeek Harness (DSH) is designed to address pain points in routing and experience accumulation in multi-agent systems. It supports creating and managing role-based subagents, using an evolution mechanism to maintain `prefercmd.md` and `memory.md`, thereby accumulating reusable commands and experience while saving tokens. The plugin supports binding and routing, strict validation, workspace isolation, and `spawn`/`fork` execution modes, and provides security isolation mechanisms. Installation requires DSH `0.1.5-rc.3` and Node.js `>=24.2.0`. Users can configure Markdown binding files to define roles, utilize built-in roles (such as the code reviewer) and `/evo-route-doctor`, among others.

dsh-plugin-wecom-bot: Integrate DSH Agent via WeCom Long-Lived Connection

The DeepSeek Harness plugin `dsh-plugin-wecom-bot` provides an interaction method based on WeCom long-lived connections, addressing the pain points of traditional setups that require public URLs, encryption and decryption, and allowlists. It supports streaming replies, zero public-network dependencies, session management, and forwarding of permission approvals. Users must install dependencies and configure the BotID and Secret, then establish a connection via WebSocket. The plugin supports both `per-user` and `per-message` modes, allowing users to switch working directories, manage sessions, and approve permissions by replying to commands directly in WeCom. Known limitations include connection count limits and unsupported message types.

dsh-tent-choice: A Parameter-Based Tent Selection Tool

The dsh-tent-choice plugin for DeepSeek Harness is a local tool designed to meet the complex calculation needs of outdoor gear selection. Built in pure Node.js, it works offline and is lightweight and controllable, providing agents with fast and accurate tent selection recommendations. Install it using `dsh plugin add github:uckkk/dsh-tent-choice`; once installed, you can invoke the `tent_choice` tool in conversations. Before use, ensure that the required dependencies are installed, and it is recommended to review the source code to verify its security.

dsh-ev-vs-fuel: EV vs. Fuel Cost Comparison Plugin

`dsh-ev-vs-fuel` is an online utility plugin in the DeepSeek Harness (DSH) ecosystem, maintained by uckkk. It is implemented in pure Node.js, requires no network requests, and is designed to accurately compare the operating costs of fuel vehicles and electric vehicles. The installation command is `dsh plugin add github:uckkk/dsh-ev-vs-fuel`. When using it, specific dependency versions must be satisfied, and it runs with the permissions of the current process. It is recommended to review the source code and license before use.

dsh-single-instance-guard: DSH Single-Instance Lock Plugin

This plugin aims to prevent session log corruption in DeepSeek Harness (DSH) caused by concurrent startup. By creating an atomic exclusive lock file (`.dsh-server.lock`) in the `DSH_HOME` directory, it ensures that only one server instance runs in that directory. If a conflict is detected at startup, the plugin either aborts with an error or cleans up an expired lock based on the liveness of the running process. The lock file is automatically removed upon process exit. With zero dependencies and support for CLI installation and manual configuration, the plugin effectively prevents log corruption errors such as "seq gap".

DSH Plugin: Cloud Well-Architected Six Pillars Review

The DSH plugin dsh-cloud-well-architected is designed to support Well-Architected Framework reviews for cloud architectures, covering the six pillars, including reliability and cost. It enables the automated integration of practices such as SLOs and Golden Signals and supports plugin-based extensibility. Users can embed it into Agent workflows, but note that specific installation commands or usage examples are not currently provided.

@koanjan/dsh-plugin-boundary: DSH Plugin Development Boundary Capability Specification

`dsh-plugin-boundary` is a boundary skill plugin for plugin development in DeepSeek Harness (DSH), maintained by KoanJan. It packages DSH boundary rules for data channels, file systems, lifecycles, and more into skills, helping developers clearly understand what they can and cannot do and avoid blind trial and error. Developers can install it from local paths or npm using the `dsh plugin` command, and modify its content by directly editing the relevant Markdown files and JS logic. The plugin is based on DSH 0.1.0-rc.6 and is licensed under the MIT License.

DeepSeek Harness Plugin: ds-notify-plugin (Windows Desktop Notifications)

The DeepSeek Harness plugin `ds-notify-plugin` is designed specifically for Windows users to receive completion status notifications during conversation generation. It relies only on built-in system tools, has no third-party dependencies, and is mounted via the Bundle mechanism, so upgrades do not break existing files. Core features include monitoring only the main conversation, displaying conversation titles, merging notifications with 1.5-second debouncing, and compatibility with PowerShell AUMID display. To install, run `dsh plugin --profile web add ds-notify-plugin` and restart; to uninstall, run the corresponding removal command.

dsh-pet-sprite: A Playable Pixel Companion Plugin for DeepSeek Harness

dsh-pet-sprite is a Web UI pet companion plugin for DeepSeek Harness (DSH) that transforms the chat interface into a platform for both entertainment and pet nurturing. It includes pet hatching and customization, conversational interactions, platformer gameplay, status-based reactions, and a token-driven growth system. Pet behavior is synchronized in real time with the model configured in DSH, and users can take control via keyboard or mouse. The plugin is telemetry-free and released under the MIT License, making it suitable for developers who want to enhance the fun of their conversations.

dsh-image-search: Multi-Engine Reverse Image Search Aggregation Plugin

The DeepSeek Harness plugin `dsh-image-search` addresses the lack of aggregated tools for reverse image search. By constructing URLs, it unifies nine mainstream engines, including Google Lens and Baidu, into a single interface and supports both JSON and Markdown output formats. The plugin uses a zero-dependency, ESM-based architecture and can run offline, with engines grouped into “General” and “Illustration/Anime” categories. Agents can call its core functions using natural language or JSON parameters to perform image provenance tracing and multi-engine search.

dsh-okf-memory: Session Memory and OKF Knowledge Consolidation Plugin

`dsh-okf-memory` is a long-term memory plugin within the DeepSeek Harness (DSH) ecosystem, designed to address the difficulty of reusing context after an agent session ends. Based on the OKF v0.1 specification, it automatically distills high-value content from conversations into standardized Markdown documents, enabling automatic recall and retrieval across sessions. Core features include a neural self-learning mechanism that dynamically adjusts memory weights based on user feedback, a technology-selection memory function that records candidate tables and decisions across dimensions such as frontend and backend, and memory graph visualization (Web-only), which allows users to view nodes, connections, and weight information. The default memory repository is located at `~/.dsh/memory/`. After installation, the plugin injects a “memory discipline” system prompt, requiring no build scripts and ready to use out of the box.

DeepSeek Harness Web Control and Pet Supply Plugin

The `w2831120407/dsh-ctl` plugin for DeepSeek Harness (DSH) integrates system control and pet interaction features into the Web settings panel. It allows users to restart or stop the dsh process with a single click from the web interface and to add dried small fish for the built-in pet “Whale Girl” without terminal operations. Installation requires linking the local directory and restarting the service; when using it, pay attention to process matching and operation confirmation.

DeepSeek Harness Plugin: dsh-washer-buy (Washer Buying Guide)

`dsh-washer-buy` is an online tool plugin for DeepSeek Harness (DSH), maintained by uckkk. By encapsulating the logic for selecting washing machines, it provides agents with a standardized `washer_buy` tool entry point and can return purchasing recommendations based on parameters without relying on external network resources. Installation requires only a single command. The plugin is implemented in pure Node.js and depends on specific versions of `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools`. It must be run with the permissions of the current process; please review the source code for security before use.

dsh-vaccine-boost: Vaccine Boost Plugin

The DeepSeek Harness plugin dsh-vaccine-boost is designed to address the lack of standardized responses when models provide vaccine advice. Maintained by developer uckkk, the plugin is built in pure Node mode with no network dependencies, ensuring stable operation. Its core function is to return clear vaccine booster recommendations based on input parameters. Developers can install it using the specified command and directly invoke the tool registered by the plugin during a session. Since the plugin runs with the permissions of the current process, it is recommended to review its source code and license before use. This plugin provides deterministic support for agents in health recommendation scenarios.

dsh-title-length: Title Length Suggestion Tool

dsh-title-length is a pure Node.js plugin for DeepSeek Harness (DSH), maintained by uckkk. It focuses on providing precise, network-independent title word-count calculation and suggestion capabilities, with `title_length` as its core tool. Users should ensure that the environment meets the required dependency versions, install it via the command, and invoke the tool during a session to obtain word-count analysis results. It is suitable for integration into agent workflows that require automatic title-length validation.

DeepSeek Harness Plugin: dsh-punycode

DeepSeek Harness (DSH) provides a plugin named `dsh-punycode` for Punycode conversion of Internationalized Domain Names (IDNs). The plugin is implemented in pure Node.js, has no network dependencies, and supports both encoding and decoding. Users can install it with the `dsh plugin add` command and invoke its tools directly within DSH sessions. This plugin provides foundational IDN processing capabilities for the DSH ecosystem and is suitable for agent development scenarios.

dsh-prune-guide: A Plugin That Returns Plant Care Plans Based on Parameters

The DeepSeek Harness plugin `dsh-prune-guide` is a lightweight plant-care plugin with no network dependencies. It provides the `prune_guide` tool, which can generate a local plant-care guide based on the specified parameters. After installation via the terminal, users can invoke the tool in a session to retrieve a guide. The plugin is built with Node.js, requires specific dependencies, and runs with the permissions of the current process.

DeepSeek Harness Plugin: dsh-order-threshold for Threshold-Based Discounts

The `dsh-order-threshold` plugin for DeepSeek Harness (DSH) is designed to address numerical calculation needs in conversational agent development. Maintained by `uckkk`, it focuses on quickly processing add-on order threshold discount calculations. Implemented purely in Node.js, it supports precise calculations based on parameters and returns detailed results, running efficiently without an internet connection. Users can install it using the official command and invoke the tool directly within a conversation, making it suitable for scenarios that require immediate amount calculations in the dialogue flow.

dsh-mover-pick: DeepSeek Harness Migration Solution Plugin

`dsh-mover-pick` is a local workflow plugin under the DeepSeek Harness (DSH) ecosystem, provided by the maintainer uckkk. It focuses on generating moving and relocation plans based on input parameters, is fully implemented in Node.js, does not rely on any network requests, and supports offline execution. The installation command is `dsh plugin add github:uckkk/dsh-mover-pick`, and the tool name after installation is `mover_pick`. This plugin is suitable for scenarios that require dynamically obtaining moving logic within DSH workflows while prioritizing lightweight usage. Before use, the source code logic and license should be verified.

dsh-knot2kph: Convert Knots to Kilometers per Hour

dsh-knot2kph is a DeepSeek Harness (DSH) plugin designed specifically to address the unit conversion requirement from knots to kilometers per hour in maritime or aviation scenarios. The plugin is implemented in Node.js, has no network dependencies, and can be installed with the command `dsh plugin add github:uckkk/dsh-knot2kph`.

dsh-habit-anchor: Habit Stacking Plugin

The DSH Harness plugin `dsh-habit-anchor` is designed to address the need for a "habit stacking" ritual in agent configuration. Maintained by developer uckkk, its core features include a pure Node.js implementation, no network dependencies, and support for offline generation of habit suggestions returned based on parameters. The installation command is `dsh plugin add github:uckkk/dsh-habit-anchor`. The plugin depends on the `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools` environments. It is suitable for developers who need locally generated habit suggestions and provides a lightweight solution for the DSH ecosystem.

DeepSeek Harness Plugin: uckkk/dsh-float-setup (Getting Started with Float Rigging)

DeepSeek Harness (DSH) plugin `uckkk/dsh-float-setup` is a pure Node.js tool for fishing float adjustment. It directly returns practical float-tuning solutions based on parameter input, with no network dependency. Users can install it using the `dsh plugin add` command, provided that specific dependency requirements are met. The plugin provides specific niche-feature support for DSH and is released under the MIT license.

dsh-fabric-care: Fabric Care Plugin

`dsh-fabric-care` is a plugin for DeepSeek Harness (DSH), maintained by istone, that focuses on providing fabric care advice. The plugin returns clear care guidance based on input parameters, requires no network connectivity, and is suitable for embedding helpful lifestyle tips or offering specific operational suggestions in conversations. Developers can install it via official commands; it runs with process-level permissions, and it is recommended to review its source code and license before use.

dsh-eye-makeup: Return Daily Makeup Options Based on Parameters

The DeepSeek Harness plugin `dsh-eye-makeup` focuses on providing eye makeup assistance solutions. It is released under the MIT License, implemented in pure Node.js, and has no network dependencies. The tool can automatically generate daily makeup suggestions based on input parameters. The installation command is `dsh plugin add github:uckkk/dsh-eye-makeup`, and it depends on `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools`. Once installed, it can be used by invoking it directly.

dsh-ear-clean: Pet Ear Cleansing and Care Plan Generator

The DeepSeek Harness plugin `dsh-ear-clean` focuses on pet ear-cleaning care scenarios. By registering a tool named `ear_clean`, it quickly generates care plans based on the provided parameters. It is implemented purely in Node.js, has no network dependencies, and is open-sourced under the MIT license. Users can install it by running a command in the terminal and invoke it directly in a session to simulate a pet care workflow.

dsh-dev-paradigms: A Programming Paradigm Query Tool

dsh-dev-paradigms is a DSH plugin maintained by uckkk, designed to address the maintenance challenges of hard-coded programming paradigm knowledge in agents. Focused on local data management, it enables quick retrieval of programming paradigm entries without network requests. Its core features include `list_paradigm` for listing entries and `get_paradigm` for retrieving details. It is suitable as a knowledge-base component for agents, while noting its MIT license and current process permission requirements.

dsh-data-use: A Precise Calculation and Recommendation Tool for Traffic Management

Within the DeepSeek Harness (DSH) ecosystem, `dsh-data-use` is a network utility plugin maintained by uckkk, designed to address the pain points of traffic calculation and parameter management in agent development. The plugin provides lightweight traffic management capabilities, supports precise calculations based on parameters, and returns clear recommendations. Its core functionality focuses on traffic management, with a fully Node.js-based implementation and no network dependency. Users can install it using a specified command, making it suitable for scenarios that require fine-grained traffic control.

dsh-cupping: Parameter-Based Coffee Brewing Recipe Generation

The DeepSeek Harness plugin dsh-cupping is a lightweight tool for generating coffee brewing plans based on parameters. It is categorized under "Memory" and features a pure Node.js implementation with no network dependencies. After developers install it via a command, they can invoke its registered tools within a session. When using it, note that the plugin runs with the permissions of the current DSH process and depends on specific versions of `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools`.

dsh-conflict-deal: A Conflict Resolution Tool That Returns Structured Methods by Scenario

The DeepSeek Harness (DSH) plugin `dsh-conflict-deal` is designed to resolve logic, input, and configuration conflicts in workflow development through a structured approach. Maintained by the developer, it provides the `conflict_deal` tool, which returns structured resolution plans based on specific scenarios. It is implemented in pure Node.js, has no network dependencies, and is licensed under the MIT License. Developers can enable the tool via an installation command and call it directly in a DSH session, thereby simplifying the conflict handling process.

dsh-coffee-ratio: DeepSeek Harness Coffee-to-Water Ratio Calculator Plugin

The DeepSeek Harness (DSH) plugin `dsh-coffee-ratio` is a lightweight tool implemented in pure Node.js, focused on calculating coffee-to-water ratios without network dependencies. It registers the `coffee_ratio` tool, supporting precise calculation based on parameters and returning results along with empirical references. Users can install it via a terminal command and invoke it directly within a DSH session. The tool is suitable for integration in local computation scenarios; pay attention to plugin permissions and the license when using it.

dsh-clamp: Numerical Clamping

DeepSeek Harness (DSH) `dsh-clamp` is a modular plugin focused on numerical processing, designed to constrain model outputs through value clamping and prevent anomalous numerical values from affecting downstream logic. It is maintained by uckkk, implemented in pure Node, has no network dependencies, and runs efficiently. Installation requires the specified version requirements of `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools` to be met. After installation, users can directly invoke the tools it provides within a session. It is recommended to review the source code and the MIT license before use, and to ensure the environment satisfies the dependencies.

dsh-claim: Claims Workflow: Return Family Insurance Planning Schemes Based on Parameters

The `dsh-claim` plugin provided by DeepSeek Harness (DSH) is a claims tool implemented purely in Node.js and maintained by uckkk. It focuses on processing insurance claims workflows and generating household insurance planning solutions, depending on `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools`. Users can install it with the `dsh plugin add` command, leveraging its local computation capabilities to reduce network dependencies, making it suitable for development scenarios that require automated processing of insurance claim logic.

dsh-carbon: Carbon Element Data Query Plugin

`dsh-carbon` is a pure Node.js plugin for DeepSeek Harness (DSH), maintained by uckkk, for locally retrieving structured data about the carbon element. It offers two methods: `element_info` for retrieving full details and `element_field` for querying specific fields. To install, run `dsh plugin add github:uckkk/dsh-carbon`; before using it, ensure dependency versions are correct. Since the plugin runs with the current process permissions, it is recommended to review its source code before installation.

dsh-car-loan: Car Loan Monthly Payment Calculator

`dsh-car-loan` is an open-source plugin for DeepSeek Harness (DSH) that performs precise local calculation of monthly car loan payments. It is implemented purely in Node.js, has no network dependencies, and requires the `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools` dependencies. Install it with `dsh plugin add github:uckkk/dsh-car-loan`. After installation, simply call the `car_loan` tool to perform the calculation.

dsh-cape-verde: Cape Verde National Information Lookup Tool

The DeepSeek Harness (DSH) plugin dsh-cape-verde embodies the "everything is a plugin" philosophy and is used for locally querying information about Cape Verde. Maintained by uckkk, it registers the `country_info` tool to retrieve basic information such as Cape Verde's capital, population, currency, and continent without network requests. Installation requires `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools`, and it is implemented in pure Node, making it suitable for offline or lightweight environments.

uckkk/dsh-budget-plan: 50-30-20 Budget Calculator

DeepSeek Harness (DSH) provides a plugin that enforces the 50-30-20 budgeting rule, aiming to address the hallucination and accuracy issues that may arise when models perform financial calculations directly. Maintained by uckkk, the plugin supports local execution in pure Node.js. Users can install it with the terminal command `dsh plugin add github:uckkk/dsh-budget-plan`. After installation, the plugin registers tools in the current DSH process, enabling sessions to invoke them directly. It is suitable for agent scenarios that require quickly generating financial allocation plans.

dsh-blame-avoid: Introducing the Blame-Avoidance Plugin

`dsh-blame-avoid` is a lightweight, network-independent DeepSeek Harness (DSH) plugin designed to simplify development workflows. It provides a “blame avoidance” tool that returns specific structured results based on parameters. The plugin is maintained by user `uckkk` and requires the dependencies `@deepseek-ai/cordis` (^4.0.1) and `@deepseek-ai/dsh-tools` (>=0.1.0-rc.6). Users can install it via a command and invoke the tool directly in a session to handle parameter differences and return structured data.

DeepSeek Harness Plugin: dsh-agriculture-terms

The DeepSeek Harness (DSH) plugin `dsh-agriculture-terms` provides local querying capabilities for an agricultural terms knowledge base, supporting on-demand retrieval. Implemented in pure Node.js with no network dependencies, it is suitable for privacy-sensitive scenarios. By registering the `list_agriculture` and `get_agriculture` tools, developers can invoke them directly in conversations, addressing the maintenance challenges of hard-coded entries. The code is open source under the MIT license.

DeepSeek Harness Standalone Session Plugin: Workspace-Free Mode

The DSH plugin `dsh-plugin-no-workspace` addresses the pain point of DeepSeek Harness forcing conversations into groups by treating workspace-free conversations as first-class citizens. The plugin achieves compatibility by wrapping the native slot and supports conversation persistence, moving conversations out of a workspace without data loss, and data security. It does not create fake folders, preserves native sidebar functionality, and supports multilingual environments. Users can install the plugin via a command and restart DSH to use it, making it suitable for scenarios that require independent conversations. The plugin is compatible with DSH version 0.1.5-rc.2.

DSH Plugin: Track Driver Development

The Track-driven development feature of the DeepSeek Harness (DSH) plugin aims to improve agent development efficiency. By integrating specifications, execution plans, and quality gates into the development process, it enhances the consistency and controllability of the workflow. The plugin supports a Track-driven development mode, inspired by Agents Conductor. Since installation and detailed usage rely on the GitHub repository documentation, users are advised to consult it directly. When using the plugin, review the source code for security and verify license compliance.

dsh-prompt-engineering: DeepSeek Harness Prompt Engineering Patterns

The `dsh-prompt-engineering` plugin, developed by `satan9394`, is an extension for DeepSeek Harness designed to streamline prompt creation. It provides standardized prompt engineering patterns, supports CoT, few-shot examples, and template reuse, and helps improve debugging efficiency. Since no official installation command is provided, users should refer to the community directory or GitHub for guidance. During installation, proper permissions and security compliance must be considered.

@qizhen2021/dsh-plugin-vision: A Vision Analysis Tool for DeepSeek Harness

DeepSeek Harness (DSH) enables text-only models to gain visual processing capabilities through the plugin `@qizhen2021/dsh-plugin-vision`. The plugin registers a `see` tool that provides three analysis channels: offline OCR (supporting macOS language detection), ASCII layout maps, and semantic descriptions (based on the `mimo-v2.5` model). Installation requires modifying `package.json` via Profile Bundles and configuring parameters such as the key path and timeout in `cordis.patch.yml`. When invoking the `see` tool, parameters can be specified to select the processing channel, and a structured JSON report is returned. Note that OCR is limited to macOS, large images require preprocessing, and the plugin's first

DeepSeek Harness Plugin: Bilibili Audio/Video Download Assistant

DeepSeek Harness Plugin: Bilibili Media Download Assistant, designed to automate downloading videos from Bilibili. It integrates the bilidown CLI and supports direct downloading within a session via a link, BV ID, or command. Its core features include anti-scraping evasion, QR code login to unlock HD quality, direct near-instant downloads via the official API, automatic selection of the widely compatible H.264 codec, batch downloading of segments and collections, and online preview. Installation requires running the specified command and depends on ffmpeg. Please note that guest mode has quality limitations, bulk operations must comply with Bilibili's risk-control policies, and the plugin does not bypass copyright protection.

dsh-plugin-android-apk: Directly Build Android Projects into APKs

The DeepSeek Harness (DSH) plugin “DSH-plugin-android-apk” allows developers to build Android projects and generate APKs directly within a DSH session. It supports automatic detection of Gradle projects, downloading the required JDK and Android toolchain (without polluting the user directory), handling non-ASCII paths, and provides a caching mechanism to improve build efficiency. Installation requires pnpm in the DSH environment. The plugin can be invoked through natural language instructions or code function calls, allowing parameters such as build variant and version to be specified. It relies on the network to download dependencies; if issues arise, check the environment configuration.

dsh-paper-slides: A Plugin for Converting Papers into Academic Presentations

`dsh-paper-slides` is a DeepSeek Harness plugin designed to address the lack of logical coherence in academic presentation generation caused by relying on abstracts. By reading the full text of a paper, it automatically sets the duration and budget for different scenarios (such as group meetings, conferences, and thesis defenses), splits the content into slides and a speaking script, and generates presentations in HTML or PPTX format. Acting as a coordinator that relies on the `dsh-ai4scholar` and `dsh-slides` plugins, it provides users with a standardized workflow from a paper to a complete academic presentation.

dsh-aloof: Integrating with the Aloof Team Knowledge Base and Approval Workflow

`dsh-aloof` is a plugin for DeepSeek Harness (DSH) that integrates Aloof's data permissions, approval workflows, and audit capabilities into local AI agents. It is maintained by gaochonggeng and licensed under the MIT license. Users must configure a token via the `ALOOF_TOKEN` environment variable; the token is bound to an Aloof address and supports data querying but does not have approval permissions. Core features include a team knowledge base (with support for document read/write and version control) and office approvals (including template lookup, form filling, and workflow submission). All write operations are controlled by two gates—local and server-side—and require user confirmation or compliance with whitelist rules. Configuration files cannot be modified directly; only YAML overrides are supported. The plugin's limitations are the lack of a dedicated UI and no support for file transfer or attachments.

DSH Smart Model Router: A Configurable Intelligent Model Routing Plugin for DeepSeek Harness

`dreamzhongju/dsh-smart-model-router` is a DeepSeek Harness plugin that simplifies multi-model integration and switching through automation. It provides intelligent classification and routing, directs requests to predefined configurations, and performs sequential failover when the primary model is unavailable. It supports reading configurations from local directories, does not store credentials, and takes over model selection in non-manual modes. Easy to install, it is ideal for users who need fine-grained control over model invocation policies.

DeepSeek Harness Plugin: dsh-plugin-novelai

`dsh-plugin-novelai` is a plugin for DeepSeek Harness that directly calls NovelAI V4.5 Full for image generation. It addresses the pain point of manually writing prompts by enabling automatic image generation from conversational descriptions, tag queries, and prompt inference. Users must configure a NovelAI token, which can be provided via chat, a configuration file, or environment variables. The plugin includes a built-in decoder for data processing, and the token is kept secure without being written to disk.

Union Lotto Assistant DSH Plugin: 1000-Draw Trend Tracking / Smart Prediction / Banker & Drag Random Number Selection

The Double Color Ball analysis plugin for DeepSeek Harness offers three core features: trend tracking, intelligent prediction, and banker-drag number selection. It automatically retrieves the most recent 1,000 draws, tracks hot and cold number status, and uses built-in prediction models to generate probability rankings, supporting rule-compliant number selection. Users can install it via the command line and invoke the `ssq` tool in a conversation to perform trend analysis, model-based prediction, or generate banker-drag numbers. Note that each Double Color Ball draw is an independent random event; this tool is for entertainment purposes only, and its data comes from official and mirror APIs.

DeepSeek Harness Plugin: satan9394/dsh-startup-business-analyst

The “Startup Business Analysis” plugin in the DeepSeek Harness (DSH) ecosystem is maintained by satan9394 and inspired by the popular project wshobton/agents. Focused on startup scenarios, it provides features for structuring business logic, estimating market size, and building financial models, aiming to simplify the preparation of business plans. As specific installation commands and usage examples are not currently available, it is recommended to consult the official directory for guidance. When using the plugin, pay attention to permissions, license information, and source code inspection.

DeepSeek Harness Machine-Level Token Statistics Plugin: dsh-token

The DeepSeek Harness (DSH) plugin `dsh-token` aims to address the complexity of tracking Token consumption in multi-session scenarios. By consolidating session logs from all `dsh` home directories, it provides a centralized Token statistics view, including totals, a heatmap, streak records, and per-model breakdowns. The plugin supports `dsh web` and reads local files for read-only statistics. Configuration options include scanning additional directories, toggling compressed summary statistics, and setting the refresh frequency. It does not include conversational costs or provide cost estimates, but accurately tracks cache reads and writes and distinguishes message types, while also disclosing detailed data such as billing coverage and retry steps.

skill-injector-plugin: Automatic Skill Injection for DeepSeek Harness

The DeepSeek Harness plugin skill-injector-plugin, maintained by Zenjibad, is designed to improve the efficiency of repeatedly and manually activating custom skills across multiple sessions. It supports automatically injecting user-selected skills either at session startup or with each request, offering two modes: "each-prompt" and "start-only". It also features a real-time skill registry, subagent support, and persistence across restarts. After installation, no configuration file is required; skills can be enabled by simply selecting them in the interface, greatly simplifying the persona-switching workflow.

DeepSeek Harness Plugin: dsh-vision-plugin

The dsh-vision-plugin for DeepSeek Harness resolves the cumbersome issue of manually describing images in multimodal conversations by automatically generating a description of image content before a message is sent and injecting it into the message. The plugin supports uploading or pasting images and uses DashScope vision models or Windows offline OCR to parse them. Users can select an OCR-only or vision-based description mode in the chat box; after a successful description, the system automatically inserts the description at the beginning of the message. The plugin follows a strict routing policy, processes images in memory as base64, and never writes them to disk. This tool simplifies multimodal interaction workflows and is suitable for developers.

DeepSeek Harness Plugin: dsh-gpt-tool-compat

The DSH plugin `dsh-gpt-tool-compat` is designed to address validation failures caused by third-party model clients appending privilege-escalation parameters in tool calls. During the `tools/pre-execute` phase, the plugin intercepts and strips invalid parameters that would inevitably be rejected (such as empty justification or unpaired fields), while preserving genuinely valid privilege-escalation requests. With fault-tolerance and exception-protection mechanisms, it implements a compatibility layer without bypassing security authorization or affecting normal calls. Installation supports Bundle or direct insertion of a Cordis Patch.

dsh-memory-forget — A Forgetting Engine for AI Agents

dsh-memory-forget is a forgetting engine plugin for AI agents, designed to transform “forgetting” from a failure into a design. It manages the memory lifecycle through TTL, decay, eviction, and auditing mechanisms to solve the context pollution problem. Core features include a pluggable memory bus, Ebbinghaus-curve-based decay, importance management (sliding update mechanism), hard injection budget limits, privacy-first physical deletion policies, and multi-agent scope support. The plugin is maintained by XIAOke8698, released under the MIT License, and has zero dependencies. It supports restoring forgotten memories and provides token accounting. Installation requires Node.js >= 20, and a StoreAdapter must be implemented to ensure data persistence.

dsh-preview: A Front-End Self-Check Tool for DeepSeek Harness

`dsh-preview` is a workflow plugin for DeepSeek Harness (DSH) that addresses the issue of agents determining task completion solely based on code while ignoring actual rendering outcomes. The plugin provides headless browser tools and a `frontend-verify` self-check skill, requiring agents to perform a complete closed-loop verification process: opening pages, checking the console, reading layout information, running interaction tests, and recording screenshots. Core features include six browser tools and an automated verification loop, ensuring that agents detect and fix errors before declaring a task complete. The plugin supports multiple browsers and includes a secure access-control mechanism that requires explicit authorization for external hosts. Through this rigorous verification discipline, `dsh-preview` transforms frontend acceptance from a manual task into a built-in agent capability, reducing the need for human intervention.

dsh-openrouter-provider-badge: Display the Actual OpenRouter Provider in DSH Session Stats Line

The `dsh-openrouter-provider-badge` plugin for DSH uses a client-side patch to display the dynamic provider actually selected by OpenRouter above the session stats line, rather than only the configured route. Installation requires three steps: install the plugin, run the patch script, and configure request headers. Although this plugin addresses the need to observe actual routing, it is no longer maintained and is only compatible with the OpenRouter channel.

DeepSeek Harness Plugin: dsh-zone-clean (Zone Cleaning)

`dsh-zone-clean` is a DeepSeek Harness (DSH) plugin maintained by uckkk. Implemented in Node.js and with no network dependencies, it returns specific home cleaning plans based on the input zone and type parameters. Developers can install it using the command `dsh plugin add github:uckkk/dsh-zone-clean`, then call the tool directly to generate cleaning suggestions. The plugin is suitable for scenarios that require adding domain-specific knowledge, such as lifestyle services, and is released under the MIT License.

dsh-wechat-et: WeChat Communication Etiquette

`dsh-wechat-et` is a DSH plugin designed to provide standardized etiquette guidelines for WeChat communication scenarios. It is implemented in pure Node.js and has no network dependencies, enabling it to instantly return relevant communication suggestions based on input parameters. After developers install it using the official command, they can invoke the related tools in a session to obtain lightweight etiquette assistance, making it suitable for developers who need dynamic suggestions and emphasize security and compliance.

dsh-vertical-org: Vertical Storage Solution Generator

`dsh-vertical-org` is a DeepSeek Harness (DSH) plugin focused on providing vertical storage solutions for small spaces to AI agents. Developer uckkk encapsulates the related logic into a `vertical_org` tool to avoid hardcoding. Once installed, it can be invoked during conversations. The plugin is implemented purely in Node, has no network dependencies, and requires dependencies such as `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools`.

dsh-sort: A Pure Node.js Array Sorting Plugin

dsh-sort is a pure Node plugin for DeepSeek Harness (DSH) that securely sorts arrays locally without network requests. Users can install it using the `dsh plugin add` command, and after enabling it, invoke the registered tool in a session to perform array sorting.

DS Plugin: dsh-solid-food (Complementary Feeding Guide)

The DSH plugin `dsh-solid-food` is specifically designed for infant and young child solid-food introduction scenarios, providing clear recommendations by registering the `solid_food` tool. It is implemented in Node.js with no network dependencies, requires installing the `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools` dependencies, and is released under the MIT license. Users can install it by running the `dsh plugin add` command. When invoking the tool, please follow medical advice.

dsh-scalp-issue: Parameter-Based Scalp Hair Loss Care Plan Generation

`dsh-scalp-issue` is a DeepSeek Harness (DSH) client plugin maintained by user uckkk, providing the `scalp_issue` tool to generate scalp hair-loss care plans based on parameters. The plugin is implemented entirely in pure Node.js and does not depend on network connections. Users can install it with `dsh plugin add github:uckkk/dsh-scalp-issue`, then invoke it in a session to obtain a care plan. It is suitable for hair-loss care scenarios; users are advised to review the source code and license before use.

dsh-pot-pick: Container Selection and Solution Generation

The DeepSeek Harness (DSH) plugin `dsh-pot-pick` provides a container selection mechanism and generates a balcony vegetable gardening plan using the `pot_pick` tool based on the provided parameters. Maintained by uckkk, the plugin is implemented purely in Node.js and requires no network connection. After installation, the tool can be invoked directly. Note that the plugin runs with the permissions of the current DSH process; please review its source code and license before use.

dsh-phone-et: Phone Etiquette Plan Generation Tool

The `dsh-phone-et` plugin provided by DeepSeek Harness (DSH) is specifically designed for telephone communication etiquette scenarios. It is implemented in pure Node.js with no network dependencies and returns appropriate etiquette guidance based on parameters via the `phone_et` tool. Installation requires the `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools` dependencies, and once enabled, it can be invoked in a session. The plugin is maintained by uckkk.

DeepSeek Harness Plugin Introduction: dsh-perfume-layer (Perfume Layering Pairings)

The DeepSeek Harness (DSH) plugin `dsh-perfume-layer` is a playful perfume-layering tool maintained by uckkk. Built with pure Node.js and requiring no external network dependencies, it is released under the MIT license. After installation, you can invoke the `perfume_layer` tool in a conversation to receive fragrance suggestions based on the supplied parameters. Designed for lighthearted interactions, this plugin provides a lightweight implementation reference for the DSH platform's "everything is a plugin" philosophy.

DeepSeek Harness Plugin: dsh-nail-issue (Nail Issue Handling)

The DeepSeek Harness (DSH) plugin `uckkk/dsh-nail-issue` provides domain-specific tools for agents to handle nails (problem nails). The plugin has no network dependencies and is implemented in pure Node.js. It registers a `nail_issue` tool that returns nail care solutions based on the provided parameters. Users can install it by running `dsh plugin add github:uckkk/dsh-nail-issue`. The tool is suitable for agents that need to integrate nail care capabilities. It requires `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools` to run and is distributed under the MIT license.

dsh-motor-current: Motor Current Calculation

The DeepSeek Harness (DSH) plugin dsh-motor-current is dedicated to local motor current calculations. It is implemented in pure Node.js and does not rely on network access. Through the `motor_current` tool, the plugin performs precise calculations based on input parameters and returns the results. It can be installed by running `dsh plugin add github:uckkk/dsh-motor-current` and is automatically registered after installation. The plugin is intended for engineering numerical computation scenarios, depends on specific versions of `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools`, runs with the current process permissions, and is licensed under the MIT License.

dsh-mode: Mode

dsh-mode is an open-source plugin for DeepSeek Harness (DSH), maintained by uckkk. It focuses on model inference and is released under the MIT License. Developed with Node.js, the plugin requires no network access and is suitable for local deployment. After installation, it registers the relevant tools, which developers can call directly within a session. Before use, ensure that the environment meets the required dependencies, and note that the plugin runs with the permissions of the current process.

DeepSeek Harness Plugin: dsh-mobius (Möbius Function)

`dsh-mobius` is a pure Node.js implementation plugin for DeepSeek Harness (DSH), centered around the “Möbius function.” It has no network dependencies and provides local session interaction capabilities. Installation requires the `cordis` and `dsh-tools` dependencies, and it can be loaded via `dsh.bundle.patch` configuration. The plugin is suitable for network-sensitive scenarios or local-only resource environments, and requires attention to permissions and the source code license during runtime.

uckkk/dsh-lip-pick: Lipstick Selection and Everyday Makeup Routines

Built on the "everything is a plugin" concept, DeepSeek Harness (DSH) recommends developers use the `uckkk/dsh-lip-pick` plugin for fun dress-up experiences. The plugin provides lipstick selection and everyday makeup routine features. Installation requires running `dsh plugin add github:uckkk/dsh-lip-pick` and depends on `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools`. The plugin is implemented purely in Node.js, has no network dependencies, and is licensed under the MIT License, making it suitable for playful interactions or scenario simulation.

dsh-laundry-temp: Laundry Water Temperature Plugin

`dsh-laundry-temp` is a community plugin for DeepSeek Harness (DSH) that focuses on providing water temperature recommendations for laundry. The plugin returns clear action plans based on input parameters, is implemented purely in Node, and has no network dependencies. The installation command is `dsh plugin add github:uckkk/dsh-laundry-temp`, after which the `laundry_temp` tool can be used. It is suitable for agent scenarios that require handling laundry temperature recommendations and can help agents make accurate decisions in related situations.

dsh-krypton: A DSH Plugin Implemented in Pure Node with No Network Dependencies

dsh-krypton is a pure Node.js DeepSeek Harness plugin designed to provide lightweight, low-level capability extensions with no network dependencies. It is developed solely using standard Node.js features, does not make external requests, and offers good runtime environment isolation, making it suitable for scenarios with strict network restrictions. Before use, ensure that core components such as `@deepseek-ai/cordis` (^4.0.1) and `@deepseek-ai/dsh-tools` (>=0.1.0-rc.6) are installed locally. The plugin runs with the current DSH process permissions; we recommend reviewing the source code to verify security.

dsh-kiribati: Kiribati Country Information Query Plugin

The DSH plugin `dsh-kiribati`, maintained by user `istone`, provides static country information lookup for Kiribati, including its capital, population, currency, and continent. Built purely in Node with no network dependencies, it can reduce connectivity costs and latency. Installation requires an environment with dependencies such as `@deepseek-ai/cordis`. It is suitable for scenarios with restricted network access or a need for cost control, and is useful for quickly obtaining basic data.

dsh-index-live: Live Online Broad Market Index Quotes

The DSH plugin `dsh-index-live` is designed to provide real-time overall market index data support for agents. It retrieves the latest market data over the network, is implemented in pure Node.js, and requires no additional network-related dependencies. It must be installed using the official command; after installation, the tool can be invoked directly within a session to assist agents with analysis and decision-making. Before use, ensure the required environment dependencies are satisfied, and review the source code to verify its security.

DeepSeek Harness Plugin Introduction: dsh-indent Text Indentation

The `dsh-indent` plugin for DeepSeek Harness (DSH) is designed to address text formatting issues in agent development and conversation sessions. It provides indentation functionality by registering a tool, is implemented in pure Node.js, and has no network dependencies. Users can install it via the official command and directly invoke the registered tool within a DSH session. The plugin is released under the MIT license and is suitable for scenarios that require text formatting.

DeepSeek Harness Heat Index Plugin

The DeepSeek Harness (DSH) plugin `dsh-heat-index` is designed to provide local heat index calculations. It is implemented entirely in Node.js, requires no network dependencies, and supports precise calculation based on temperature and humidity parameters. Developers can install it using the `dsh plugin add` command for use in restricted network environments. It relies on no external APIs, runs with DSH process permissions, and adds lightweight computational capabilities to the DSH ecosystem.

dsh-hanfu-style: Choosing Hanfu Styles

DeepSeek Harness (DSH) is designed to extend agent capabilities through plugins. This article introduces the `dsh-hanfu-style` plugin, maintained by uckkk, which generates hanfu outfit suggestions based on input parameters. The plugin is implemented in Node.js and has no network dependencies. After developers install it by running the `dsh plugin add` command, they can invoke the `hanfu_style` tool in conversations. When using it, they should check the source code and license to ensure safety.

dsh-hanfu-size: Hanfu Size Calculation Plugin

The dsh-hanfu-size plugin in the DSH ecosystem, maintained by uckkk, focuses on Hanfu size calculation and outfit plan generation. Built entirely in Node.js, it does not rely on external network requests, with all calculations performed locally. After developers install it using `dsh plugin add github:uckkk/dsh-hanfu-size`, they can invoke the registered tool `hanfu_size`, passing parameters such as height and weight to obtain accurate results. Proper installation of environment dependencies is required for use. This plugin adds vertical-domain calculation capabilities to the DSH plugin ecosystem.

dsh-go-rules: Go Rules Plugin

The `dsh-go-rules` plugin for DeepSeek Harness (DSH) provides agents with lightweight Go rules processing capabilities and supports returning Go learning plans based on parameters. The plugin is implemented in pure Node, has no network dependencies, and can be invoked in a session after installation. Note that it runs with the permissions of the current process. It is recommended to review the source code and license before installation.

DeepSeek Harness Plugin: dsh-gear-list

The `dsh-gear-list` plugin for DeepSeek Harness (DSH), maintained by uckkk, provides a “beginner gear list” feature. It can return a practical fishing plan based on the input parameters. Its core features include a pure Node.js implementation, no network dependencies, and execution with the permissions of the current DSH process. Users can install it using the official command and then use it by invoking the registered tool in a session. Given its open-source nature, users are advised to review the source code and MIT license before installation.

dsh-fx-live: Real-Time Exchange Rate Conversion

`dsh-fx-live` is a web-enabled tool plugin for DeepSeek Harness, maintained by developer `uckkk`. It addresses the issue that agents lack real-time exchange rate data when handling currency conversions by fetching the latest rates online and using them for conversion. The plugin is open-sourced under the MIT License, is easy to install, and can be invoked directly in a session without any configuration. Because it is implemented in pure Node.js, it is recommended to review the source code for security before deploying it in a production environment.

dsh-future-metaverse: Local Metaverse Entry Management Plugin

`dsh-future-metaverse` is a Node.js-based DeepSeek Harness (DSH) plugin designed to provide localized metaverse entry data management capabilities. The plugin requires no network dependencies, interacts with DSH via CLI, and supports listing and querying entries by ID or name. It is suitable for use on intranets or in restricted environments, providing a lightweight data management solution for DSH.

dsh-flash-gn: Flash Guide Number Calculator

DeepSeek Harness (DSH) provides a localized flash GN value calculation tool plugin, `dsh-flash-gn`. The plugin is maintained by community developer `uckkk`, implemented in pure Node, and has no network dependencies. Through the `flash_gn` tool, it allows agents to calculate flash settings precisely based on the provided parameters. Users can install and enable the plugin via the command line and invoke the related tools in DSH sessions. The plugin runs with the current process permissions; checking the source code and license before installation is recommended.

dsh-fish-pick: Ornamental Fish Selection and Aquarium Care Plan Generation

`dsh-fish-pick` is a pure Node.js, network-independent fun plugin for DeepSeek Harness (DSH), maintained by uckkk. The plugin provides ornamental fish selection and aquarium maintenance solutions. It depends on `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools`, and its tools can be invoked directly in conversations after installation. It is suitable for scenarios requiring ornamental fish Q&A or care advice, is licensed under the MIT License, and users are advised to check the source code for security before use.

dsh-fertilizer-npk: Fertilizer NPK Ratio

DeepSeek Harness (DSH) provides the `dsh-fertilizer-npk` plugin for calculating fertilizer NPK ratios. Maintained by uckkk, it returns clear fertilizer blending recommendations based on crop or soil requirements. It supports parameterized inputs to generate specific recommendations, is implemented in pure Node.js, has no network dependencies, and is suitable for local or offline deployment. Users can install it via a command and invoke it within a session, making it suitable for use cases that require specialized agricultural calculation capabilities.

dsh-espresso: Espresso Extraction

The DeepSeek Harness (DSH) plugin dsh-espresso is a pure Node.js tool focused on espresso extraction. It can generate a structured extraction protocol offline based on parameters including coffee dose and water temperature. The plugin aims to reduce reliance on LLM inference and improve output determinism, making it suitable for agent development that requires professional coffee guidance. The installation command is `dsh plugin add github:uckkk/dsh-espresso`. Because it runs with the current process permissions, it is recommended to review the source code before use to ensure security and compliance.

dsh-coupon-opt: DeepSeek Harness Coupon Decision Tool

dsh-coupon-opt is a DeepSeek Harness plugin maintained by uckkk, designed to provide local coupon calculation capabilities. Implemented in pure Node.js with no network dependencies, it supports precise calculations based on parameters and returns detailed results. Developers can install the plugin via the terminal and directly invoke its utility functions within a session to extend business logic. The plugin has specific dependency requirements, and reviewing its source code before use is recommended. It is suitable for running in restricted network environments, meeting lightweight computation needs without accessing external services.

dsh-contract-unlimited: Indefinite Contract Detection

DeepSeek Harness provides the `dsh-contract-unlimited` plugin, which encapsulates the logic for determining open-ended employment contracts. The plugin accepts parameters for precise calculation, returns detailed results and statutory basis, and supports automated processing of labor contract determinations. The installation command is `dsh plugin add github:uckkk/dsh-contract-unlimited`. Implemented entirely in Node.js, it has no network dependencies at runtime. Before use, it is recommended to review the source code and license to ensure compliance.

dsh-breakeven: Break-Even Calculation Plugin

The dsh-breakeven plugin for DeepSeek Harness (DSH) is designed for break-even calculations in small business operations. Built with pure Node.js and free of network dependencies, it rapidly calculates the break-even point based on input parameters. Developers can install it using a terminal command and invoke its tool directly within a DSH session to obtain the results, making it suitable for scenarios requiring quick business data analysis.

dsh-beer-pick: Beer Selection

The `dsh-beer-pick` plugin for DeepSeek Harness (DSH) is designed to address specific logic issues in agent decision-making. By using parameterized suggestions and running locally, it reduces development costs. Maintained by uckkk and implemented in Node.js, it has no network dependencies. It requires the `@deepseek-ai/cordis` and `@deepseek-ai/dsh-tools` dependencies to install, and once installed, its registered tools can be used directly in a conversation.

dsh-barking-fix: Barking Correction

`dsh-barking-fix` is a DeepSeek Harness plugin maintained by uckkk, guided by the core philosophy of “everything is a plugin.” It falls under the “Barking Correction” category and aims to provide clear correction suggestions for non-standard model outputs through a parameterized approach. Implemented purely in Node with no network dependencies, it can be installed and then used in a session to call the `barking_fix` tool for quick validation and error correction, making it suitable for scenarios that require manual intervention.

dsh-baozi-steam: Steaming Time

`dsh-baozi-steam` is a DeepSeek Harness plugin maintained by uckkk, categorized under “Fun Dress-Up.” It is a local tool implemented purely in Node.js that calculates steaming plans for wheat-based dishes based on parameters without requiring internet access. Easy to install, secure, and open source, it is well suited for network-sensitive environments.

DeepSeek Harness Plugin: dsh-arithmetic-series (Arithmetic Series)

`dsh-arithmetic-series` is a DeepSeek Harness (DSH) plugin maintained by uckkk, designed to add mathematical capabilities to sessions. It is implemented in pure Node.js, encapsulating the logic for generating arithmetic sequences into a tool without depending on external APIs. Once installed via `dsh plugin add github:uckkk/dsh-arithmetic-series`, the relevant tool becomes automatically available in sessions, ensuring stable calculation results.

dsh-appliance-life: Precise Appliance Lifespan Calculation Plugin

The `dsh-appliance-life` plugin for DeepSeek Harness (DSH), maintained by uckkk, focuses on accurate appliance lifespan calculations and recommendations. Built entirely with Node.js, it has no network dependencies. It depends on a specific version of DSH, enabling developers to install it via the command line and invoke tools directly within a session to assess appliance usage status.

dsh-amount-calc: Hot Pot Dining Party Portion Calculator Plugin

The DeepSeek Harness plugin `dsh-amount-calc` is a calculation tool focused on hot pot dining scenarios. It follows the "everything is a plugin" philosophy, is maintained by uckkk, and aims to address logic calculation needs in specific scenarios. The plugin is implemented purely in Node, requires no network dependencies, and supports returning hot pot dining plans based on parameters. Developers need to install it using the `dsh plugin add` command; once enabled, registered tools can be invoked in sessions. The plugin runs with the permissions of the current process, making it suitable for developers who need a lightweight calculation solution.

dsh-mermaid-web: Mermaid Chart Rendering Plugin for the DeepSeek Harness Web UI

dsh-mermaid-web is a DeepSeek Harness Web UI plugin designed to render Mermaid code blocks in Markdown into interactive SVG diagrams. It provides zoom controls, a full-screen workspace, Hot Module Replacement (HMR), and keyboard shortcut support. The plugin relies on the DOM structure for rendering and supports error handling as well as a strict security mode. Developers can install it directly using a pnpm command without a build step.

DeepSeek Harness Plugin Marketplace

tasihi89/dsh-plugin-market is the official plugin marketplace for DeepSeek Harness, designed to address the challenge of plugin discovery. It offers features such as a trending panel, in-app installation, and a bilingual interface, allowing users to browse and install popular plugins directly within the DSH client. The tool lowers the barrier to obtaining plugins, helping users find inspiration or quickly install commonly used plugins. Note that plugins run with the permissions of the current process, so it is recommended to review the source code and license information before installation.

dsh-snake: DeepSeek Harness Web Hungry Whale Mini-Game Plugin

The `dsh-snake` plugin for DeepSeek Harness (DSH) renders a Greedy Whale mini-game as a background on the development page. The blue whale acts as a full-viewport background element with click-through behavior, so it does not interfere with content interaction, while developers can control the whale to eat golden tokens. The plugin features a minimal control capsule, direction keys for steering, the spacebar to start/pause, data persistence, and light/dark theme support. The installation command is `dsh plugin --profile web add github:Solicide/dsh-snake`, and the service must be restarted after modifying the code. This tool is intended to provide subtle entertainment breaks and is released under the MIT license.

DeepSeek Harness Plugin: dsh-workflow-loop (Workflow Loop Perspective and Specification)

`satan9394/dsh-workflow-loop` is a skill plugin for DeepSeek Harness (DSH), guided by the core philosophy that “everything is a plugin.” It aims to address the difficulty of transforming loop logic in complex workflows into delegable specification definitions. Maintained by developer satan9394, the plugin focuses on providing a loop-oriented perspective and specification definitions for workflow design. Since specific features, installation commands, and usage examples have not yet been published, users can consult its GitHub repository or skill catalog page for source code and license information.

DSH Plugin: Theme Factory

DSH Theme Factory is a "fun dress-up" plugin maintained by user satan9394, designed to address inconsistent visual styling across the DeepSeek Harness (DSH) skill interface. It provides preset theme configurations to quickly adjust the appearance without modifying core code. The plugin is released under the Apache-2.0 license and can be obtained via a directory page or its GitHub repository. Due to missing documentation, no specific usage instructions are currently available; developers should inspect the source code and permissions before installation.

DeepSeek Harness Plugin: dsh-incremental

The DeepSeek Harness plugin dsh-incremental addresses the pain points of excessively large change scopes and difficult rollbacks in agent development. Its core design philosophy is “everything is a plugin,” and through an incremental implementation strategy, it helps developers keep the system functional by making small, incremental commits rather than one-off large-scale changes. Developers are required to visit the source code repository to review installation instructions and specific usage examples. Because the plugin runs with the permissions of the current DSH process, its source code logic must be reviewed before use to ensure compliance with security policies and licensing requirements.

DeepSeek Harness Plugin: dsh-idea-refine (Idea Refinement Tool)

sat9394/dsh-idea-refine is a DeepSeek Harness workflow plugin that uses a divergent-convergent approach to turn vague ideas into actionable plans. Inspired by agent-skills, it runs with the current process permissions. Review the source code and license before use. See the GitHub repository for more details.

DeepSeek Harness (DSH) Plugin: dsh-diagram-design (Editorial-Grade Diagram Design)

`dsh-diagram-design` is a DeepSeek Harness plugin maintained by user `satan9394`, designed to address the need to manually write SVG code or call external services when agents output diagrams. The plugin provides 38 self-contained SVG diagram types and supports editor-level diagram design. It runs with the privileges of the current process and is suitable for scenarios where agents need to generate structured diagrams. For installation instructions, please refer to its GitHub repository or plugin directory page. Before use, users should review the source code and license agreement.

DeepSeek Harness Plugin: CSS Artistic Style Recipe

This article introduces a DeepSeek Harness (DSH) plugin named "CSS Art Style," designed to address the issue that generic style instructions lack specificity. Maintained by user `satan9394`, the plugin provides directly callable CSS aesthetic recipes, including styles such as liquid glass. Users can obtain the source code via the GitHub repository or the plugin directory page. The plugin runs with the permissions of the current process and is suitable for developers who need to apply CSS visual aesthetics, but the source code should be inspected before installation to ensure safety.

dsh-career-ops: Job Search Command Center

The `dsh-career-ops` plugin for DeepSeek Harness (DSH) is designed to centralize and standardize job-seeking workflows through a modular architecture. Positioned as a “career command center,” it provides core features such as job description evaluation, resume optimization, and interview preparation, aiming to reduce the complexity of manual operations. Since no explicit installation command is provided officially and documentation retrieval failed, users should independently review the source code and license before installation.

DeepSeek Harness Plugin: dsh-academic-research Academic Research Pipeline

This article introduces a DeepSeek Harness (DSH) plugin designed specifically for academic research. Inspired by well-known projects, it aims to automate the entire workflow—from defining research questions and searching and verifying literature, to synthesizing findings, assisting with writing, and conducting final review—through an “everything is a plugin” approach. The plugin supports research pipelines and helps improve academic efficiency. Users should pay attention to permission isolation and license compliance; see its GitHub repository for installation and usage details.

dsh-smart-router: DeepSeek Harness Smart Model Routing Plugin

The `dsh-smart-router` plugin implements intelligent routing via a virtual model, automatically switching to configured models based on request complexity and visual requirements without manual intervention. Core features include difficulty-based routing (hard, normal, and easy tiers), automatic image detection to invoke a vision model, and two classification modes: heuristic or LLM-based. Configuration requires specifying the model for each tier and the visual mode (replacement or pass-through), and it is recommended to use models from the same provider to optimize caching. The plugin supports a fail-open fallback mechanism and can coexist with existing vision plugins.

dsh-brief-session-title: DSH Session Title Compression Plugin

The `dsh-brief-session-title` plugin for DeepSeek Harness (DSH) resolves the issue of verbose default title generation by optimizing system prompts. It fully reuses the official workflow and features zero intrusion, automatic exclusive takeover, fault tolerance, and configurability. After installation, titles for new sessions are compressed into a single character or word, improving readability and retrieval efficiency while supporting manual renaming and locking.

dsh-plugin-prune: DeepSeek Harness Plugin Usage Statistics and Cleanup

`dsh-plugin-prune` is a DeepSeek Harness plugin developed by Octo-o-o-o, designed to address the management challenges caused by accumulated “dead plugins.” The plugin operates in a purely observational mode and does not modify business state. It collects real-world tool invocation data, such as call counts, error rates, and latency, and provides a visualized health report in the “Plugin Health” tab of the Web interface. Users can use this information to score plugins and clean up unused ones. The data is persisted locally only and requires use with a specific version environment.

DeepSeek Harness Plugin dsh-subagent-vision: Enabling Pure-Text Main Agents to Read Images Directly in Conversations

The DeepSeek Harness plugin `dsh-subagent-vision` addresses the inability of pure-text models to handle images. It delegates image processing tasks to a configured vision subagent, enabling the main agent to remain text-only while receiving visual results. The plugin supports pasting, dragging and dropping, and importing images via file path; provides a native Web GUI experience; and enforces a 25 MB image size limit. After installation, users only need to describe the task to the agent, and the plugin automatically invokes the subagent's `read_image` tool to complete it.

dsh-skill-indexer: A Two-Stage Intent Retrieval Plugin for Local SKILL.md

dsh-skill-indexer is a Host Bundle plugin for DSH, designed to address the challenge agents face in precisely locating local SKILL.md files. It establishes a two-level indexing mechanism that leverages keyword coverage and IDF for intent routing, supporting a three-state fallback (hit/low/miss). The tool runs entirely locally with zero dependencies, offers both command-line and programmatic invocation, and supports hash-based incremental updates.

dsh-thrum: DeepSeek Harness Input Vibe Plugin

dsh-thrum is a playful DeepSeek Harness plugin designed to improve the input experience with visual and audio feedback. It visualizes keyboard keys, which flip to reveal animals on keypress, and includes typing flame effects, input box vibrations, and reply chimes. Users can customize the keyboard appearance, flip frequency, flame style, and vibration intensity via a settings panel or configuration file. The open-source plugin is ideal for developers looking for vivid feedback during monotonous typing.

dsh-windows-workspace-guard: Windows Workspace and Credential Protection Plugin

`dsh-windows-workspace-guard` is a Windows agent-protection plugin developed specifically for DeepSeek Harness (DSH), designed to control dangerous file operations by agents. Independent of official components, it intercepts tool calls and provides three decision mechanisms—`PASS`, `ASK` (approval), and `HARD BLOCK` (interception)—to prevent file deletion, credential reading, or system-state corruption. It supports multiple tool calls and includes pre-check capabilities, audit logs, and summary statistics. Installation is straightforward, and users can configure `workspaceRoots` and `sensitivePaths` to isolate environments. The plugin does not send network requests; it is a conservative boundary-control tool suitable for use in Windows environments.

dsh-plugin-discussion-intent: DSH Discussion Mode

DSH `dsh-plugin-discussion-intent` is a Discussion Mode plugin designed to prevent drift in complex AI conversations. It continuously injects user intent, boundaries, and evaluation criteria to ensure the conversation remains coherent and goal-oriented. Core features include preventing intent drift, allowing sub-agents to select their own models, deep settings that move from vague to actionable, on-the-fly course correction via a four-line Rail, and support for continuing discussions. The plugin depends on DSH 0.1.0-rc.6, supports checkpoint persistence, and can be controlled through the `/discussion` command to manage discussion depth, model, and status.

dsh-daily-ledger: A Local Daily Check-in and Ledger for DeepSeek Harness

`dsh-daily-ledger` is a local ledger and check-in plugin for DeepSeek Harness (DSH). It runs fully offline and does not upload data. Core features include daily check-ins, token and cost statistics (with support for daily, model-level, and itemized breakdowns), and a Web floating window for viewing status and customizing the background. The plugin automatically listens to DSH events to record usage, and supports configuring price snapshots and custom time zones. Installation is simple: you can check in or query statistics through chat commands, providing users with a convenient tool for local cost tracking and habit management.

dsh-deepsea: Turn Session Context into Deep-Sea Slacking

`dsh-deepsea` is a web plugin for DeepSeek Harness that transforms dialogue context pressure into a deep-sea fishing game. The hook depth changes in real time according to context length. The game features four water layers, five oceans, and independent creature pools and holographic card generation. Cards are generated by the MiniMax model and their textures are baked using Python. Caught fish can be released into a fish pond world. The plugin requires setting up a MiniMax API key and a Python environment, supports cloud synchronization of game results, and is designed to intuitively demonstrate AI processing difficulty through gamification.

DeepSeek Harness Plugin: dsh-plugin-subagent-delete

DeepSeek Harness (DSH) lacks an interface for deleting subagents, causing their list to accumulate. The `dsh-plugin-subagent-delete` plugin fills this gap by providing tools and HTTP routes, including `list_subagents`, `delete_subagent` (for permanent deletion), and `release_subagent` (to release and restore subagents). It supports recursive operations, ownership validation, and automatic Web UI refresh, and is compatible with DSH 0.1.0-rc.8.

dsh-memory: A Codex-Style Persistent Memory Plugin for DeepSeek Harness

The DeepSeek Harness plugin `dsh-memory` provides persistent memory for agents, addressing the loss of context in multi-turn conversations. Its core features include persistent storage, automatic summary injection, a memory toolset, and configuration via a web interface. Installation requires specifying a version hash and restarting the environment. The plugin operates in the `$DSH_HOME/memories/` directory and achieves long-term memory capabilities similar to Codex by automatically distilling conversation turns and periodically merging them into a global summary. It supports initializing seed memory from `AGENTS.md` and provides tools for reading and writing memory.

DeepSeek Harness (DSH) Plugin: dsh-event-watch

The DeepSeek Harness (DSH) plugin dsh-event-watch aims to address the inefficiency of waking idle Sessions in agent development. It supports listening for file system events and GitHub events, waking a target Session and sending a Prompt when specific conditions are triggered, and includes a cold-start feature. The plugin runs with the permissions of the current DSH process and is suitable for automated handling of external events such as code commits or configuration changes.

DeepSeek Harness Branch-Aware Conversation Difference Plugin

`dsh-fork-diff` is a read-only branch comparison plugin designed specifically for DeepSeek Harness, aimed at solving the difficulty of comparing long responses and tool trajectories after forking. It automatically discovers parent, child, and sibling branches, provides a side-by-side view that highlights differences in responses, tool calls, and results, and supports two-column desktop or single-column mobile layouts. The plugin is read-only throughout and does not modify any files, while offering pagination, quick alignment mode, and filtering capabilities. Users can select branches to compare from the session title bar, eliminating the need to manually process complex histories and enabling efficient identification of points of divergence.

DeepSeek Harness Aggregated Web Search Plugin

The `dsh-web-search-aggregation` plugin unifies multiple search providers—including AnySearch, TinyFish, and Tavily—into a single entry point. It enhances availability through a priority queue and an ordered fallback mechanism, supporting multi-key pool rotation, budget-aware timeouts, and real-time configuration. This plugin addresses the issue of a single search service becoming unavailable and is well suited for high-availability web search scenarios involving DeepSeek Harness agents.

dsh-jj-vcs: Jujutsu Version Control Plugin for DeepSeek Harness

The DeepSeek Harness (DSH) plugin `dsh-jj-vcs` addresses version-control bottlenecks in multi-agent collaboration by introducing the Jujutsu version control system. It provides each agent with an isolated workspace and leverages a change-based model to deliver clear attribution auditing, concurrency safety, and operation tracing. Its core features include creating independent changes, explicit permission tiers, controlled publishing, and Windows compatibility. Developers must install it manually and depend on JJ 0.44.0, using the coordinator script to start tasks and verify changes, thereby providing a robust version-management solution at the community level for enterprise-grade multi-agent projects.

dsh-conversation-threads: DSH Dialog Message Bar Navigation Track

`dsh-conversation-threads` is a plugin that adds a Codex-style navigation thread to DSH, helping users quickly locate historical messages in long conversations. The thread appears on the right side of the sidebar, includes short lines corresponding to user messages, and supports hover previews, click-to-jump, and scroll synchronization. The plugin can be installed persistently as a Bundle or activated dynamically as a Skill. It is implemented entirely on the client side and poses no privacy risks.

dsh-premise-guard-cn: A Context-Premise Loss Prevention Plugin for Chinese Novel Writing

`dsh-premise-guard-cn` is a Chinese-optimized version of the `ICCuse/dsh-premise-guard` plugin, designed to address the loss of critical premises caused by long-context compression in DeepSeek Harness. It includes built-in automatic anchor-extraction rules for Chinese (such as Chinese quotation marks, terminology chains, and paths), supports manually maintaining an anchor list, and provides the `premise_anchor` tool and `/anchor` command for management. After context compression, the plugin automatically detects and injects warnings, prompting the model to verify any lost settings and effectively preventing logical collapse in long-form Chinese creative writing.

dsh-memory: Cross-Session Memory and Vector Retrieval Plugin for DeepSeek Harness

`dsh-memory` is a cross-session memory plugin designed for DeepSeek Harness, aimed at resolving the loss of context after agent restarts. Maintained by developer zhouzhencheng07, it features core capabilities including automatic background memory, maintaining daily workspace notes, memory retrieval, and optional vector search. The plugin runs with the permissions of the current process; users are advised to review its source code and license before installation. It provides an effective solution for handling long-term memory in the DSH ecosystem.

dsh-sxs-news-collector: A DSH Plugin Aggregating Hot Topics from 4 Sources

`dsh-sxs-news-collector` is an open-source plugin for DeepSeek Harness (DSH) that aggregates trending topics across the web for content creation and data analysis. Developed by SXS, it integrates four hot-search sources—Baidu, Toutiao, Douyin, and Weibo—and provides one-click fetching of multi-source data, deduplication, and rate-limit protection. Users can install it via pip or add it from a local path, invoke it using the `sxs_news_hot` function, and filter results by source or limit the number of items. The plugin is free and open source, making it suitable for developers who need real-time trending topics.

dsh-screenshot: A WeChat-Style Screenshot Plugin for DeepSeek Harness

dsh-screenshot is a screenshot plugin built specifically for DeepSeek Harness Web, designed to streamline debugging and demonstration workflows. It integrates into the toolbar and supports two modes: “Capture Screen Area” and “Capture with Windows Hidden.” It uses local PowerShell to crop and process images, eliminating data uploads to protect privacy and security. The plugin also automatically injects image paths, enabling a model to read image content via MCP Vision. It requires Windows and DSH 0.1.x, and is well suited for users who need to perform local debugging or demos.

DeepSeek Harness Plugin: dsh-plugin-guardian Safe Uninstall and Rollback

dsh-plugin-guardian is a plugin within the DeepSeek Harness ecosystem, designed to address the accumulation of residual files caused by plugin uninstallation. It provides features such as safe uninstallation, snapshot rollback, leftover file cleanup, health checks, and visual operation interfaces, eliminating the need to rely on command-line usage. Users can obtain installation and usage instructions through the skills catalog or the GitHub repository. It is suitable for scenarios that require fine-grained plugin management; before using it, users should review the source code and license information.

dsh-restart-web: One-click safe restart of processes in the DSH Web settings panel.

# Introduction If you use DSH via the web interface, you have likely encountered moments where the page freezes or becomes unresponsive, or where you need to restart the process for a newly installed plugin or modified configuration to take effect. The conventional approach is to return to the terminal, find the process occupying port 3080 and terminate it, then manually start DSH, and finally refresh the browser page yourself. `dsh-restart-web` streamlines this workflow into a single button in the settings panel: clicking it safely restarts the entire DSH process, and the browser automatically refreshes and reconnects after approximately 5 seconds, allowing you to stay within the web interface the entire time. DSH follows the philosophy that "everything is a plugin," and such interface-level utilities are ideally suited to be implemented as plugins. The following sections introduce its features, installation methods, and working principles. # What is dsh-restart

dsh-web-search-toggle: Add a "Search-then-Answer" Web Search toggle for DSH sessions

**Introduction:** When developing agents using DeepSeek Harness (DSH), a common issue arises: models typically do not have internet access by default, or they only invoke `web_search` when they deem it necessary. If you require the responses in a specific session to include the latest web information, you typically have to repeatedly emphasize "search before answering" in every round of prompts. DSH's philosophy is "everything is a plugin," and such behaviors can be handled by plugins. The dsh-web-search-toggle introduced below is exactly such a plugin: it adds a "Web Search" toggle switch to the chat input toolbar. Once enabled, it forces the agent to call `web_search` before answering in every round of that session, following the official DeepSeek approach.

satan9394/dsh-codebase-design: A DSH plugin that requires verification before installation.

Preface: The philosophy of DSH is "everything is a plugin"; capabilities are accessed via plugins, and the community directory allows browsing and searching by entry. However, the quality of information in the directory entries varies: some plugins come with complete READMEs, installation commands, and examples, while others have only a one-line description, or documentation cannot even be found. This article introduces the latter: `satan9394/dsh-codebase-design`. As of the writing of this article (2026-09-07), its GitHub README scraping has failed, and verifiable information is extremely limited. Therefore, the focus of this article is not to list features, but to explain how to verify information in this situation before deciding whether to install. What is this? Currently, only the following points can be confirmed: - Repository URL: `https://gith`

what-was-that: A small octopus to help you understand the terminology in AI responses.

# Introduction When using AI assistants to write code or conduct research, their responses often contain terms like `RAG`, `LoRA`, and `MCP`. If you don't recognize them, you have to interrupt the flow: either copy them to search elsewhere, or ask follow-up questions in the chat—the former disrupts the rhythm, while the latter pollutes the context. Here is an introduction to a small plugin that automates this: "what-was-that". It adds a little octopus desktop pet to the DSH web interface, automatically scans for terms in AI replies, and explains them in plain language. # What is it? what-was-that is a DSH static client plugin developed by SH-9999 (running on the web platform), MIT licensed, and published to npm (`what-was-that@0.1...`).

DSH Plugin Introduction: satan9394/dsh-cad-modeling

# Preface For DSH-based agent development, the fundamental way to extend capabilities is installing plugins—the philosophy of DSH is "everything is a plugin". However, when selecting plugins, we often encounter a situation where the repository is accessible, but the README cannot be retrieved, and the description, installation instructions, and license information are nowhere to be found. satan9394/dsh-cad-modeling is just such a repository. Below, I will introduce the information currently confirmed, content not yet verified, and the evaluation steps suggested when documentation is incomplete. The information was last verified on September 7, 2026. # What is it satan9394/dsh-cad-modeling is a DSH plugin repository listed in the community directory, maintained by GitHub user satan9394. Directory page details...

dsh-rules: DeepSeek Harness glob activation rules plugin

# Introduction Writing collaboration rules for agents often involves stacking all conventions into a single global prompt. However, when there are many rules, two issues arise: irrelevant content consumes the context window, and the model struggles to distinguish which rule applies to which files. Claude Code addresses this using `rules.md` and `# Path:` sections—rules declare the file scope they govern themselves and only take effect when touching matching files. If you are developing on DeepSeek Harness (DSH) and want this same mechanism, the `dsh-rules` introduced below provides exactly that: rules match files using glob patterns, and when the agent reads or edits matching files, the rule content is automatically injected into the conversation. # What is it dsh-rules is created by rj-jiangyich

dsh-scientific-research: Research Workflow Skills Package for DSH

dsh-scientific-research is a DSH scientific research scenario skill pack maintained by satan9394, categorized under the workflow category, with the philosophy of "Everything is a plugin."

dsh-minigame-plugin: Enable DeepSeek Harness to directly generate playable HTML5 mini-games

dsh-minigame-plugin is a DeepSeek Harness plugin developed by sasajun2345 and licensed under the MIT License. It codifies mini-game development experience into skill documentation. Users describe requirements directly in conversation, and the plugin automatically handles engine selection and pattern reuse, outputting a single-file, zero...

dsh-sampling-sliders: Add a real-time adjustable sampling parameters panel for DSH

In agent development, this situation often arises: the default temperature or maxTokens of the model is unsuitable, and one wishes to temporarily adjust them to compare results. However, this usually requires modifying the configuration file and restarting the service every time. If you only want to adjust the temperature from its default value to 0.2 for a single conversation, current solutions either require modifying global settings that affect all sessions or lack an interface to make changes. The philosophy behind DeepSeek Harness (DSH) is "everything is a plugin," making this kind of requirement a perfect fit for a plugin. The dsh-sampling-sliders plugin introduced below addresses this: it adds a "Sampling" button to the input bar. Upon clicking, sliders appear to adjust the temperature and maxTokens for each individual session.

dsh-decision-map: Visualizing session execution trajectories into intuitive decision maps.

# Preface When running agent tasks with DSH, a session often contains multiple rounds of reasoning and dozens of tool calls, resulting in long, sequential event logs. The built-in "Trajectory" view is ledger-style—it is queryable, but not intuitive for quickly answering questions like "what happened in this round," "which step was the slowest," or "where were the tokens spent." DSH's philosophy is that everything is a plugin, and the session view itself leaves an interface for integration. The dsh-decision-map introduced below is a layer of visualization created by leveraging this interface. # What is it dsh-decision-map (Decision Map) is a DeepSeek Harness bundle plugin maintained by Scitiger-AI, current version 0.1.0, MIT license. One-sentence positioning: visualizing the execution trajectory of the current session...

dsh-diary-detective (Cyber Diary Detective): A warm mirror reflecting your diary.

# Preface When building agents using frameworks like DSH (DeepSeek Harness) that adopt the "everything is a plugin" philosophy, you must implement long-term memory yourself: deciding where to store it, how to compress it, and how to retrieve it next time—every step involves manual intervention. Diaries represent a typical scenario: they require archiving the original text while ensuring the model can grasp key points when reviewing them the next day, without starting with a cold, statistical report. Here, we introduce dsh-diary-detective (Cyber Diary Detective), which packages diary I/O, memory compression, and gentle feedback into a single plugin. The project is maintained by SageChenAI and is licensed under the MIT License. # What is it One-sentence positioning: A warm mirror that reads diaries, reveals blind spots, and provides gentle feedback. It performs two distinct layers of functions: 1. Mechanical Layer: Storing spoken diaries as files and reading back the original text

oh-my-ticket: DSH's 5-Level Task System and run Batch Execution Plugin

oh-my-ticket (OMT) is a task management plugin designed for DeepSeek Harness. It treats the five-level task hierarchy (Epic→Story→SubStory→Ticket→SubTicket) as first-class citizens, addressing the issue of task lists scattered across conversations that are difficult to persist. It utilizes a dual-write architecture with SQLite for metadata storage and Markdown for content storage. It supports committing to the project directory via Git and allows for a global home rollback, alongside the `om t_reindex` command to rebuild the index. The UI provides three container types: a left sidebar, a floating window, and tabs. It features tree interaction, search and filtering, two-way linkage (mentions, automatic display of related tickets), and SSE-based real-time refreshes. The core `run` batch execution mechanism snapshots tickets into ordered batches, passing through...

dsh-mobile: Enable mobile phones to connect to DeepSeek Harness via LAN

# Introduction Daily operations for DeepSeek Harness (DSH) typically occur on the desktop, whether via the web GUI in the browser or the desktop shell window. When users step away from their desks, there is generally no ready-made entry point to check the conversation history of a specific workspace or send a follow-up prompt on their mobile phones. dsh-mobile solves this problem. It establishes a mobile connection bridge within the local area network (LAN): the desktop displays a QR code for the mobile device to scan and pair, and once the desktop approves the connection, the mobile device directly opens a DSH client optimized for mobile use. The following sections introduce the positioning, features, installation, and usage of this plugin. # What is this dsh-mobile is a plugin for the DeepSeek Harness core, maintained by s17179XTY

dsh-token-usage: Add a Token Usage Analysis Page to DSH Web GUI

dsh-token-usage is a Token usage analysis plugin for DSH Web GUI (npm package @samecorner/dsh-client-ui-token-usage, v0.1.3, MIT licensed). It adds a "Token Usage" tab at the top of the session, rendering projection data such as `tokenUsage` and `contextPressure` from the host as visual panels. Features include per-round KPI cards, context pressure progress bars (warning at 80%), context composition, and per-round bar charts.

dsh-archive-manager: View, restore, and permanently delete archive sessions in the DSH Web GUI

# Introduction If you use the DeepSeek Harness (DSH) Web GUI daily, you might have encountered this situation: after a session is archived, it can no longer be found in the sidebar. Although the backend does maintain a global `archivedSessionIds` collection, the official UI provides no entry point to view archived sessions, unarchive them, let alone permanently delete a session. Simply hiding archived sessions from the list will only render them irrecoverable at the GUI level. Furthermore, deleting a session requires sequentially cleaning up multiple interdependent stores—transcription directories, workspace accounting, archive flags, and projection caches—otherwise, residues will be left behind. dsh-archive-manager was written to fill this gap. Below,

dsh-plugin-practice: A DSH plugin practice repository that fits into your profile

# Introduction The philosophy behind DeepSeek Harness (hereinafter referred to as DSH) is "everything is a plugin": Tools, Configurations, Services, Events, and middleware must all ultimately be integrated as plugins. For those new to plugin development, the difficulty usually lies not in the lack of documentation, but in the absence of a minimal example that ties concepts together, runs successfully within a profile, and provides observable output at every step. Below, we introduce dsh-plugin-practice. It breaks down the core concepts of plugin development into six progressive lessons. The code accumulates step-by-step as you progress through the lessons. It is itself a standard DSH Bundle that can be installed and uninstalled using official commands. What is dsh-plugin-practice

dsh-matt-pipeline: Visualize Matt Pocock Skills Workflow in DSH sidebar

# Introduction When developing agents with DSH, a common approach is to break down tasks into fixed stages: first capturing ideas, then filtering, refining through questioning, writing specifications, breaking down tickets, and finally implementing and reviewing them one by one. If each step corresponds to a skill, the user must remember the skill names and manually type `/skill` commands one by one. After completing a single cycle, seven or eight commands are memorized, and if interrupted, the user must recall where they left off. `dsh-matt-pipeline` solves this problem: it registers Matt Pocock's set of skill workflows into DSH and provides a visual pipeline panel in the sidebar. A single click dispatches the corresponding skill to the current session. # What is dsh-matt-pipeline (Version 0.2.0, MIT

dsh-more-session-operations: Unread, Copy ID, and Delete enhancements for the DSH Web session menu.

In DeepSeek Harness (DSH), the "..." menu in the session row of the Web sidebar serves as the primary entry point for daily operations. However, the actions provided by the official menu are limited: there is no "mark as unread" feature for sessions completed but viewed later; there is no direct way to copy the Session ID to scripts; and archiving executes immediately upon clicking, leaving no buffer against accidental clicks. When dealing with a large number of sessions, these minor inconveniences require extra steps every time. DSH's philosophy is "everything is a plugin," making it perfect for filling such gaps at the interface layer via a plugin. `dsh-more-session-operations` is an example of such a plugin: it adds three new actions—"Mark as Unread," "Copy Session ID," and "Delete Session"—to the "..." menu in the session row and adds a confirmation dialog for the official archive function. The following sections cover functionality, installation, and typical usage scenarios.

dsh-tool-everything: Integrate Everything full-disk file search for DSH Agent

`dsh-tool-everything` is an Everything search plugin for DSH (DeepSeek Harness) that registers the `everything_search` tool for an Agent. It directly invokes the local Everything NTFS index on Windows to perform millisecond-level full-disk searches by filename, without needing to enable an HTTP server or es.exe. The plugin loads the bundled Everything64.dll (taken from Everything SDK 1.5) via the koffi FFI, communicates with Everything through an IPC channel, and returns structured results (path, size, is a folder). It supports the full Everything search syntax, such as wildcards.

dsh-cloudflare-browser-run: Provide real browser crawling capabilities for DeepSeek Harness

dsh-cloudflare-browser-run is a DSH plugin maintained by RealAlexandreAI that integrates Cloudflare Browser Run's headless Chrome into DSH, addressing the limitation of standard HTTP scraping in handling JavaScript-rendered pages. The plugin provides three tools: browse (returns clean markdown by default, with optional screenshots or PDF), screenshot, and pdf. Its design features include real browser access and SSRF protection.

dsh-workspace-scope: DSH Plugin to Enable/Disable Skills and Global MCP Per Workspace

# Introduction Under DeepSeek Harness's (DSH) "everything is a plugin" philosophy, Skills and global MCP servers installed into the environment tend to accumulate. The issue is that these capabilities are visible to all sessions by default: every new session's startup context carries a comprehensive list of tools, while a single project often only utilizes a fraction of them. `dsh-workspace-scope` addresses this issue: it enables each workspace (project) to determine which Skills and global MCPs to expose to new sessions. The result is similar to installing numerous language extensions in VS Code, where only the necessary ones are enabled for each specific project. Below, we introduce its positioning, core features, installation method, and specific usage. # What is dsh-workspace

dsh-thinking-summary: Adding a Chinese summary to the DSH Web thinking process

# Introduction When running agent tasks on DSH's Web interface, the model's native state is just a line reading "Deep diving," and the reasoning process itself is in English. For users who cannot read English reasoning, it is difficult to keep up with the model's pace; by the time they figure out what the model intends to do, the command may have already started executing. dsh-thinking-summary addresses exactly this issue. DSH's philosophy is "everything is a plugin." This plugin displays the next step plan in real-time to the right of "Deep diving" and inserts a Chinese summary card below each block of reasoning. The following sections introduce its features, installation methods, and configuration. # What is it dsh-thinking-summary is a DSH Web plugin created by Rosma

dsh-swarm-router: DSH swarm plugin for routing heterogeneous tasks to the most suitable model

Introduction: In agent development, you often encounter task batches like this: one involving 17x23 arithmetic problems, another requiring code writing and prime number verification, and a third being long-form translation. Sending them all to the same model queues simple tasks on flagship models, wasting time and tokens; assigning difficult tasks to cheap models often leads to errors. Manually selecting a model for each task becomes unmaintainable as the workload increases. dsh-swarm-router addresses this problem: it organizes a batch of heterogeneous tasks into a "Sub-Agent Matrix Swarm"—tasks are rows, and candidate models are columns. The router selects one cell for each row, and then uses DSH's `ctx.subagents` to turn each cell into a process-bound sub-agent parallelized to the selected model. # What is dsh-swarm-router

dsh-better-at: Move DSH Web's @ mention index to the browser

# Introduction In the DeepSeek Harness (DSH) web interface, the `@` symbol serves as the primary entry point for referencing files and sessions. Under the default implementation, every keystroke triggers a request to the Host for candidate filtering; consequently, delays in menu opening and filtering accumulate significantly when there are numerous workspace files and historical sessions. dsh-better-at addresses this by pre-fetching file and session indices into the browser, performing all filtering and sorting locally, while preserving the native behavior of `@` references and mention formats. The following introduces this plugin. # What is dsh-better-at dsh-better-at is a plugin for the DSH Web GUI, positioned as a "local caching accelerator for `@` file/session references."

dsh-plugin-clawrouters: A plugin package and a key to connect DeepSeek Harness with Chat, Images, Videos, and Search.

# Introduction When developing agents for DeepSeek Harness (hereinafter referred to as DSH), a common pain point is scattered capabilities: chat requires a provider, while image generation, video generation, and web search each require configuring keys and writing adapters. After ClawRouters consolidates these capabilities into an OpenAI-compatible API and a single key, `dsh-plugin-clawrouters` packages this key and set of capabilities into a DSH plugin package. Once installed and configured, it can be selected directly within DSH. The following introduces the plugin's positioning, features, installation, and configuration methods. # What is this? `dsh-plugin-clawrouters` is created by rop...

DeepSeek VisionPlus: Text goes to DeepSeek, images automatically routed to free vision model pools

When using DeepSeek Harness (DSH), there is an unavoidable limitation: DeepSeek's official models are text-only and cannot process images. When screenshots, charts, or interface photos appear in a message, one must either manually transcribe the content or switch the entire conversation to another multimodal model, resulting in a break in context. DeepSeek VisionPlus takes a different approach: text requests continue to be handled by the DeepSeek official API, while images are automatically routed to a pool of free visual models, with the understanding results returned to the same conversation. DSH's philosophy is that "everything is a plugin," and this capability is implemented as a plugin. Below is an introduction to its features, installation, and daily usage.

dsh-trivium: In-process graph memory plugin for DeepSeek Harness

dsh-trivium is a process-in-memory graph memory plugin for DeepSeek Harness (DSH) based on TriviumDB, designed to solve the issue of agent memory not persisting across sessions, eliminating the need for external vector databases or independent services. It structures memory into four tools—ctx_find, ctx_read, ctx_remember, and ctx_link—with nodes categorized into Entities, Preferences, Decisions, and Experiences, while edges record associations. By default, the plugin operates quietly, injecting a compact graph of approximately 400 tokens into new sessions only once and querying the rest on demand. It supports manual editing for correction, strict write hygiene (filtering out garbage text and keys), optional embedding vector retrieval, and Git collaboration. It is installed via the `dsh plugin` command, utilizes one `.tdb` file per workspace, and fails without blocking.

dsh-theme-palettes: DeepSeek Harness Color Scheme Infrastructure

DeepSeek Harness (DSH) follows a "everything is a plugin" philosophy, yet the interface theme layer is an exception: the Appearance picker (light/dark/system) is not open to third parties, and the theme registry does not accept plugin modifications. Consequently, third-party plugins previously lacked a standard entry point for applying complete color schemes to the web interface. The `dsh-theme-palettes` introduced below adopts a different approach: rather than modifying the theme registry, it overlays palettes onto the current theme via `theme.overrideTokens`. It also provides a `themePalettes` registration API, enabling any plugin to contribute its own.

dsh-plugin-archive-manager: Adds session archive management to dsh Web UI

# Introduction dsh sessions can be archived, but there are few built-in entry points for handling post-archiving tasks. Whether you want to retrieve a specific old session, perform a bulk restore, or automatically archive inactive sessions based on rules, there is currently a lack of a visual interface for these operations. The dsh-plugin-archive-manager introduced below is designed to solve this problem by incorporating these capabilities into the dsh Web UI settings page. # What is it? dsh-plugin-archive-manager is a dsh plugin maintained by qinyre, released under the MIT license, with the current version being 0.3.0. It provides a top-level section titled "Archive Management" in the Web UI settings page, covering three functions: browsing archived sessions, unarchiving, and automatic archiving based on rules. It pre...

dsh-codex-import: Read-only Compatibility Scanner for DeepSeek Harness Codex Configuration

# Introduction If you maintain two sets of agent environments, Codex and DeepSeek Harness (DSH), the issue during configuration migration is rarely a lack of tools, but rather a lack of clarity: Codex's `config.toml`, hooks, Skills, and `AGENTS.md` are scattered across various user-level and project-level locations. Determining which can be directly mapped to DSH and which are unsupported requires manual verification of each item one by one. Copying configurations directly poses risks due to differences in permission semantics and inconsistent support scopes for hooks. `dsh-codex-import` isolates this step: it performs a read-only inventory first and provides a categorized result; version 0.1 does not execute any migration. DSH's philosophy is "Everything is a plugin", this...

dsh-plugin-installer: DSH Plugin Installation Assistant with Backup and Rollback

# Introduction The philosophy of DSH is that "everything is a plugin," and most functional extensions are handled by installing plugins. However, the act of installing plugins remains a manual task: executing installation commands, modifying configurations, and restarting processes. Once restarted, the current conversation is interrupted; the model has to start from scratch, or you must manually paste the context back. What's worse, if the newly installed plugin breaks the startup, dsh cannot start at all, requiring you to troubleshoot dependencies and restore configurations entirely on your own. dsh-plugin-installer encapsulates the entire workflow—'backup → install/uninstall → restart → auto-continue'—into a model tool, delegating execution to the model with automatic rollback in case of failure or startup anomalies. The following sections introduce its features, usage, and precautions. # What is this dsh-plugin-installer is Quop

dsh-group-chat: Start a multi-AI group chat in DSH

# Introduction The philosophy of DSH is "Everything is a Plugin," and many WebUI capabilities are injected via plugins. In agent development, a common requirement is to have several models discuss around the same context in turns—one proposes a solution, one critiques it, and one summarizes it. Without specialized tools, context usually has to be manually moved between multiple sessions, and the speaking order must be monitored manually. The `dsh-group-chat` introduced below turns this into a group chat format: you act as the "group owner" in the DSH WebUI, manage a set of configurable AI roles, and enable them to take turns speaking around a shared context. The plugin is maintained by Qx002, and the current version is 0.1.0. # What is this dsh-group-chat is a DSH native...

dsh-agent-driver-writehere: Integrate the WriteHERE long-form writing loop as DeepSeek Harness's second reasoning loop.

# Introduction Building a long-form writing agent on DeepSeek Harness (hereinafter DSH) typically involves encapsulating retrieval, drafting, and revision into a set of tools attached to the default ReAct loop. This approach squeezes the outline, draft, and revision into a single loop, leaving no dedicated space for different types of work and lacking independent handling for mid-process modifications. `dsh-agent-driver-writehere` adopts a different mounting point: instead of appending tools to the default loop, it implements the WriteHERE long-form writing loop (heterogeneous recursive planning) as an agent driver, running as a second inference loop. The following sections introduce its design, installation, and usage. # What is dsh-agent-driver

dsh-more-agent-presets: Multiple optional Agent presets for DeepSeek Harness

# Introduction The default prompts included with DeepSeek Harness (DSH) are optimized for DeepSeek models. If you use non-DeepSeek models or older model versions in DSH, the default preset interactions may not fully leverage the model's capabilities. Additionally, some developers prefer discussing and writing code simultaneously rather than letting the agent execute tasks independently. dsh-more-agent-presets addresses these issues: it provides four sets of optional Agent presets as a plugin for DSH, catering to different model adaptations and collaboration styles. Below is an introduction to the plugin's features, installation instructions, and precautions. # What is it dsh-more-agent-presets is a

dsh-plantuml: Directly Render PlantUML Diagrams in DSH Sessions

dsh-plantuml is a DSH web plugin (v0.1.0, MIT license) maintained by qschen86, used to automatically render PlantUML/puml fenced code blocks in assistant messages into...

meow-cachebilling: A DSH Plugin Embedding Real-time Billing into the Context Ring

# Introduction When running long-context tasks with DSH, cache hit rates are often impressive, but the total cost does not stop climbing: the longer the context, the more cached tokens are carried in each request, causing the bill to snowball. "A 100% cache hit rate but getting more expensive the more you use" describes exactly this situation. It is not enough to simply know the token usage; one must also know the monetary cost of these tokens to determine when to start a new session or compress the context. Below, I introduce meow-cachebilling, a DSH web plugin that puts real-time billing into a context ring overlay. # What is it? meow-cachebilling is a DSH plugin maintained by Phant0Meow and is MIT licensed.

silverhand-dsh-pet: A desktop pet at the bottom-right corner of the DSH Web GUI that reacts to agent status.

# Foreword Most DSH users encounter the same issue: agents can run for several minutes or even longer. When you switch to another window to work, your first priority upon returning is to check whether the agent is currently running, waiting for approval, or has stopped due to an error. Currently, the only method is to switch back to the Web GUI to monitor the output or check the logs. `silverhand-dsh-pet` takes a different approach: it delegates the status display to a desktop pet. It resides permanently in the bottom-right corner of the DSH Web GUI. The pet breathes when the agent is idle, goes to work when the agent is busy, and acts up when an error occurs. A quick glance at the corner tells you the status of the agent. The following sections introduce the specific features, installation steps, and troubleshooting methods of this plugin. `silverhand-dsh-pet` is a standard DSH plugin...

dsh-xiaomi-tts: Add Xiaomi MiMo TTS voice reading to DSH Web

Introduction: When running agents on DSH Web, the output is plain text by default. To enable voice output, a common approach involves implementing a custom Text-to-Speech (TTS) layer: applying for an API key, setting up a proxy on the Host side, handling audio streams in the browser, and cleaning up code blocks and URLs before playback. While each individual step is not complex, integrating them can be time-consuming. dsh-xiaomi-tts streamlines this process into a directly installable DSH Web plugin. Upon installation, an "Read Aloud" button appears next to the dialogue. Clicking this button initiates playback. What is this? dsh-xiaomi-tts is maintained by ppy-web. Its purpose is succinctly defined as: adding Xiaomi MiMo TTS voice playback to DSH Web. It is built upon Xiaomi MiMo...

dsh-llm-headers: Inject custom HTTP headers into DeepSeek Harness's LLM requests

# Introduction When deploying DeepSeek Harness (DSH) in an enterprise intranet, a common barrier arises: the egress gateway validates the request's `user-agent`, and requests lacking the specified brand text are rejected outright. However, DSH's LLM adapter forces the writing of its own attribution `user-agent`, which cannot be bypassed by simply modifying a field in the configuration file. `dsh-llm-headers` solves this problem: it is a DSH plugin that injects your configured request headers at the HTTP layer for requests sent to the LLM provider. Below is an introduction to its principles, installation methods, and typical usage. # What is dsh-llm-headers `dsh-llm-headers` is for Deep...

openma-ai/dsh-mcp-apps: Add MCP Apps support for DeepSeek Harness

# Foreword MCP Apps allow MCP tool results to carry `ui://` HTML resources, which are rendered by the client into an interactive interface. To get this working in DeepSeek Harness (DSH), a complete set of components is required, including the protocol Host, sandbox renderer, and display mode management. In contrast to existing approaches: DSH's Web build comes with an inline-only renderer that only covers a single surface type. Below, I introduce openma-ai/dsh-mcp-apps: it packages the MCP Apps Host and Web renderer into a standard Cordis plugin, filling in this layer of capability. # What is openma-ai/dsh-mcp-apps (npm

Software Project Governance: Centralize the management of AI code delivery evidence, gates, and risks

software-project-governance is an MIT open-source project maintained by peterwangze, providing an evidence-based governance layer for AI coding delivery to address the issue where facts such as task changes, reviews, and verifications are scattered across chat logs. Core features include evidence-based delivery (recording what was changed, why, how to verify, and remaining risks), gatekeeping and centralized risk management, distinguishing between genuine independent reviews and execution in degraded/dependent environments, and getting started in five minutes by initializing the .governance/ directory via /governance. The plugin supports multiple agent loading paths such as Claude Code, Codex, and Gemini CLI, and provides a read-only local web console. The default preset is classic-phase-gate.

dsh-ask-guard: Add a timeout guard for ask_user_question to prevent turns from hanging indefinitely.

# Preface If you have used `ask_user_question` in DeepSeek Harness (DSH), you may have encountered the following scenario: the agent asks a question, and the question card is pushed to the browser via a mux WebSocket in a single push. Once the page is sent to the background, the laptop enters sleep mode, or the connection enters a half-open state, that frame is lost. Without a heartbeat, the client cannot perceive the question's existence, so the tool waits indefinitely. Practical tests have observed hangs lasting over 6 hours, and the only recovery method is manually pressing "Stop". For upstream feedback, see deepseek-ai/deepseek-harness#1554 (discussion). dsh-ask-guard is designed to address this issue.

dsh-plugin-table-zoom: One-click scrollable floating window for DSH chat tables

dsh-plugin-table-zoom is a chat table floating window plugin for the DeepSeek Harness (dsh) Web GUI, maintained by Pasumao under the MIT license, currently at version 0.3.4. It addresses the issues in DSH chat flows where long and ultra-wide tables are difficult to read completely and inconvenient to copy: since version 0.1.1-rc.2, dsh has enabled the md-table-wide branch for wide tables with 4 or more columns, but the default overflow is hidden and scrollbars only appear on hover, still restricting the reading experience. The plugin automatically injects a "⛶ Float View" button for tables with $\ge$ 9 rows or horizontal overflow exceeding 2px. Clicking it opens a floating window that supports independent scrolling, drag-and-drop movement, and resizing via the bottom-right corner (from 320×200 to the viewport limit), with the title displaying the row and column counts; supports Ct...

dsh-plugin-image-tools: Three tools to add image capabilities to DSH Web GUI

# Introduction When using the Web GUI of DeepSeek Harness (DSH) to create agents, there are three image-related scenarios that have always been difficult to handle: 1. The model wants you to select an image from a set of candidates (e.g., choosing a book cover), but images cannot be placed in the tabs; the model can only print the file path for you to open manually. 2. The model generates an image and wants to mix it with text in the response body, but there is no channel for this either. 3. You send an image into the chat, and a text-only model adapter directly reports `UNSUPPORTED_CONTENT`, interrupting the turn. dsh-plugin-image-tools is a DSH plugin designed for these three scenarios: image selection card, inline images in the response, and blind model receiving images. Three tools, each handling a specific task, with zero configuration.

dsh-browser-plus: A DSH Plugin Making Agent Browser Truly Visible and Controllable

## Introduction A common practice in browser automation is to run the Agent in a headless process invisible to the user. The problem is that the user has no way to confirm what is happening on the page or where the Agent clicked; when encountering steps that require login or manual judgment, there is also no natural handover method. dsh-browser-plus is designed to address this issue. It keeps the browser window visible to the user while providing the Agent with reliable Chrome DevTools Protocol (CDP) capabilities. Users can interact directly on the real page, and the Agent can execute, take over, and resume tasks synchronously. ## What is it dsh-browser-plus is a visual Agent styled with EGO, specifically crafted for DeepSeek Harness (DSH).

dsh-tool-tariff: DSH Peak/Valley Time Period Query and Balance Reminder Plugin

# Preface If you want to schedule tasks to run during off-peak hours and finish before peak hours arrive, you need to be constantly aware of whether it is currently peak or off-peak and when the next switch will occur. Existing methods involve maintaining a schedule manually and cross-referencing it, which is cumbersome and prone to error; similarly, checking the remaining balance of the DeepSeek API requires interrupting your current work. `dsh-tool-tariff` is a DSH (DeepSeek Harness) plugin that integrates these functions into the session: the model can directly answer time-related queries, automatically reminds you before peak hours, and a real-time status badge remains in the top bar of the Web GUI. DSH's philosophy is "Everything is a plugin," making this capability suitable for mounting as a plugin. The following sections introduce the features, installation, configuration, usage, and precautions in that order. # What is this d

dsh-agents-plugins: A Bridge package for integrating Codex, Claude Code, and Pi plugins into DeepSeek Harness

# Introduction If you have previously installed plugins in Codex, Claude Code, or Pi, you likely have a collection of assets at your disposal: hooks, slash commands, MCP servers, and skills. After switching to DeepSeek Harness (DSH), these cannot be migrated directly—they are written according to the format of another agent runtime, and their loading methods and lifecycles do not match. Rewriting them one by one is clearly not cost-effective. DSH's philosophy is "everything is a plugin," and such bridging work can itself be completed by a plugin. The openma-ai/dsh-agents-plugins introduced below is such a bridge: allowing you to install Agents in DSH through a package.

dsh-port-guard: A DSH plugin for handling port occupancy by first identifying it

dsh-port-guard is a DeepSeek Harness plugin developed by PangYiMing (v0.1.0, MIT license) used to handle EADDRINUSE port conflicts when the dev server starts. It transforms the brute-force kill into an informed judgment: first using a script to locate the occupying process, viewing information such as start time, uptime, controlling terminal, and project ownership, and then handling three scenarios—if its own dev server is still running...

dsh-dracula-theme: Apply the classic Dracula dark theme to DeepSeek Harness

# Introduction When using the Web GUI of DeepSeek Harness (DSH), the visual appearance follows the client's color scheme. If you want to switch to a pleasing dark theme, the conventional method is to fork the client and manually edit the CSS. This is tedious and requires rebasing every time you upgrade. DSH's philosophy is "Everything is a plugin." The client has a built-in theme runtime, allowing color schemes to be registered as plugins without modifying the client source code. The `dsh-dracula-theme` introduced below is precisely such a theme plugin: it brings the classic Dracula dark color scheme to DSH. # What is this `dsh-dracula-theme` is maintained by ossFrankFrank and is released under the MIT License. It is based on the official Dracula...

dsh-browser-control: A CDP plugin that lets DSH agents reuse your daily Chrome login state

## Introduction Connecting browsers to AI agents often involves launching a fresh headless instance. The problem lies in maintaining login sessions: many pages require logged-in cookies to display real content, and re-authenticating in isolated environments every time is costly. However, directly enabling `--remote-debugging-port` on a regular Chrome instance is explicitly prohibited in newer versions of Chrome—the default user-data-dir does not allow the debug port to be opened. Below, I introduce `dsh-browser-control`, a DeepSeek Harness (DSH) plugin that uses the Chrome DevTools Protocol (CDP) to drive the browser. Its core selling point is...

dsh-xiangqi: DSH client plugin for playing Chinese Chess during AI thinking

dsh-xiangqi is an MIT-licensed open-source DSH Web client plugin developed by ovdoesw, allowing users to play Chinese Chess with a floating cartoon pet while waiting for the AI to think. The gameplay is purely for entertainment, fully offline, and non-blocking. It features a built-in custom TypeScript engine (negamax + alpha-beta pruning + iterative deepening, executed asynchronously with chunking) with three difficulty levels: beginner, intermediate, and advanced.

omdsh-chatmode: Add a Chat/Work mode switch to DeepSeek Harness

omdsh-chatmode is a chat mode plugin for the DeepSeek Harness web GUI (npm package `@omdsh-plugins/omdsh-chatmode`, v0.2.5, MIT) designed to solve the barrier of "asking questions on the fly without having to select a working directory first." It provides a Chat / Work toggle above the conversation: the Chat mode allows starting a dialogue without selecting a project directory; all chat...

dsh-molecule-viewer: Render interactive 3D molecular structures in DSH sessions

# Introduction For those working on protein structures or molecular docking, the typical workflow for viewing structures is: the model calculates the results, you save the PDB file, switch to PyMOL or a web tool to open it, and then take a screenshot to paste back into the discussion. Files are shuffled between the chat session and external software, requiring a context switch and interrupting your train of thought. In DSH (DeepSeek Harness), it is not difficult to have the model read a molecular file, but the challenge lies in making the structure itself appear in the session interface. The `dsh-molecule-viewer` plugin solves exactly this step: by passing a file path or molecular data, it renders a 3D molecular viewer directly in the session that supports rotation, zooming, and style switching. # What is it dsh-molecule-viewer is DS...

dsh-screenshot-diff: A DSH plugin for pixel diffing existing screenshots

# Introduction When engaging in interface development or having agents operate on pages, a question often arises: Has the page actually changed after this modification? Visually comparing two screenshots is prone to missing details, and writing a pixelmatch script from scratch every time is too repetitive. dsh-screenshot-diff is a workflow plugin within the DeepSeek Harness (DSH) ecosystem. The philosophy of DSH is "everything is a plugin," and repetitive verification actions are perfect for being encapsulated as plugins: a single command takes two existing screenshots as input and outputs a diff image with red-marked differences, a labeled tri-comparison image, and a JSON snippet that can be consumed by scripts. It does not handle screenshot capture; the input must be two screenshots you have already obtained. # What is dsh-screenshot-

dsh-batch-regression: Run the same command N times, use the median instead of a single result to draw conclusions.

# Introduction Most developers who have done performance tuning have encountered this situation: running a benchmark once yields 1200ms, while rerunning it at a different time yields 1900ms. You change a line of code to verify if it's faster; one run feels faster, but another run feels like an illusion. Single results are subject to environmental noise, so drawing conclusions directly from them is unreliable. The philosophy of DSH (DeepSeek Harness) is "everything is a plugin," and this problem has a corresponding plugin to solve it. Below, I introduce `dsh-batch-regression`, maintained by PangYiMing: it runs the same command N times, takes the median and distribution, and uses statistical analysis rather than single results to draw conclusions. # What is this? `dsh-batch-regression` is a DeepSeek Harness (DSH) plugin, positioned to...

dsh-large-proj-perf: Eliminating blocking caused by fork, history loading, and disk spilling in DSH super large sessions

**Preface** The design philosophy of DSH (DeepSeek Harness) is "everything is a plugin." Several performance-oriented plugins already exist in the community: @linxin666/dsh-perf handles observability and frontend rendering load reduction, while dsh-pref-kit performs streaming incremental merging at the source. However, there is a class of issues regarding large existing sessions that these plugins cannot address: once the number of events in a single session accumulates to several hundred thousand, forking sub-sessions, reopening historical sessions, and persisting fork results to disk all become synchronous heavy tasks on the event loop—causing anything from a few hundred milliseconds of lag to a single core running at full capacity for 20 minutes, or even a direct RangeError. This article introduces dsh-large-proj-perf (v1.2.0, MIT license, author orangeofc).

dsh-cursor-subscription: Use Cursor subscription directly in DeepSeek Harness

dsh-cursor-subscription is a DSH plugin maintained by orrinzeng (MIT licensed) that allows existing Cursor subscription users to directly use subscription models in DSH via browser PKCE login. It requires no API Key and does not depend on the Cursor IDE or CLI. Core features include local credential storage and automatic token refresh, dynamically discovering available models via GetUsableModels (falling back to a built-in list on failure), streaming conversation output, converting Cursor tool requests to local DSH tool execution, as well as usage monitoring and runtime configuration (tool round limits, HTTP retry strategies). To install, use `dsh plugin --profile web add dsh-cursor-subscription`.

omdsh-basemode: DeepSeek Harness session mode registry, switcher, and sidebar dots

# Introduction The philosophy of DSH (DeepSeek Harness) is "everything is a plugin." When applied to the web GUI, this raises an immediate issue: both the Chat mode plugin and the Code mode plugin want to occupy the session bar. Who provides the "switch mode" toggle? In the sidebar, a row of sessions all look identical. Which ones belong to which mode, and which one are you currently viewing? If the registry is released with a specific mode, other modes would have to depend on that entire mode's package, inverting the layering relationship. The omdsh-basemode introduced below achieves exactly this by extracting this "seat" from the various "modes." # What is this `@omdsh-plugins/omdsh-basemode` (Plugin Center Display Name: Base Mode / 基

dsh-bigmodel-catalog: Enable DeepSeek Harness to use Zhipu GLM Coding Plan and glm-5.3

`opensaozi/dsh-bigmodel-catalog` is a DSH (DeepSeek Harness) model catalog plugin used to integrate Zhipu BigModel / GLM Coding Plan into the DSH model selector. It is maintained by OpenSaozi under the MIT License, with the npm package name `@deepseek-ai/dsh-model-catalog-bigmodel`. Core features include unlocking the 1M context window and 128K max output for glm-5.3, and supporting glm-4.5, glm-4.6, and glm-5; it connects to the Zhipu Coding PaaS endpoint to enable real-time catalog reconciliation, with glm-5.3 retained as a verified exception. The plugin registers an independent zai adapter.

dsh-better-stats: Add a real-time cost and status bar to DSH Web UI

# Introduction When running long tasks with DSH (DeepSeek Harness), keeping track of costs is the most difficult part: DeepSeek-V4-Flash and DeepSeek-V4-Pro have different pricing, and official RMB rates vary by peak and off-peak hours. Input is charged in buckets based on cache hits or misses, and the usage of parallel sub-sessions within an agent team must be aggregated. It is hard to estimate how much the current turn has cost halfway through the streaming output, and manually reconciling costs using token details afterward is also a hassle. The philosophy behind DSH is "everything is a plugin," and information display in the Web UI can similarly be handled by plugins. dsh-better-stats does exactly this: it places a status bar directly below the Composer,

dsh-identity-control: Customize dialogue persona next to DSH input bar

dsh-identity-control is a DSH custom persona plugin maintained by GitHub user orpheus0829 (v0.1.0, MIT license). It provides a button and an edit panel next to the chat input bar, allowing users to manually input persona text and toggle it with a single click. Once enabled, the persona is automatically injected into all new conversations; disabling it restores the default without requiring a restart, and it does not override DSH safety rules. The plugin operates via official commands.

dsh-think-summary: DSH Real-time Segmented Summary Plugin for Long Thinking

# Introduction The Chain of Thought (CoT) of reasoning models is valuable for debugging and evaluation, but in chat interfaces, it usually scrolls past quickly. Once the answer is complete, looking back to see "how the model reasoned" often lacks an entry point; one can only sift through raw logs. dsh-think-summary solves exactly this problem. It is a dual-side plugin for DSH (Cordis architecture): it detects the model's long reasoning processes, segments the thinking process, and generates summaries step-by-step, displaying them in the chat area in real-time. It also comes with a settings page and historical persistence. The following sections introduce it in the order of features, installation, verification, and precautions. # What is it The plugin is named dsh-think-summary, and its repository is located at https://github.com/oakcakerolls/dsh-think-summary

dsh-superpower: Integrating 14 developer skills from obra/superpowers into DeepSeek Harness

## Introduction When building a coding Agent with DeepSeek Harness (DSH), a common challenge is the difficulty of reusing methodological capabilities: items like requirements clarification, Test-Driven Development (TDD), and systematic debugging typically exist as `SKILL.md` skill files. To use them, one must write their own loading and registration logic, while also handling injection timing and compatibility issues across different presets. Existing solutions often involve manually copying skill files into each project and rewriting the loading code from scratch. dsh-superpower turns this into a plugin: after a single command installation, the 14 skills from obra/superpowers are automatically registered into the DSH skill registry, and accompanying session discipline is injected at the appropriate times. Below, we introduce its features, installation method, and working principle.

kidai-snapshot-guard: An Automatic Snapshot and Rollback Plugin for DSH

# Preface The philosophy of DeepSeek Harness (DSH) is "everything is a plugin." The most frequent operations in daily use are installing plugins, updating them, and modifying settings. The risk lies here: once a change breaks DSH, what you need most is the state "before the change." Relying on manually copying the configuration directory before every change is both easy to forget and difficult to guarantee integrity. **kidai-snapshot-guard** automates this: it completes the "Save → Confirm → Restore" loop for snapshots within DSH. Once installed, it just works without requiring you to remember to do anything. # What is it? One-sentence definition: An automatic backup manager inside DSH — save before shutdown, verify after boot, one-click rollback. Plugin by NokorinNishikino.

dsh-auto-vision: Equip DeepSeek Harness's text-only main model with vision tools

## Introduction When developing agents with DeepSeek Harness (DSH), you are likely to encounter this issue: the built-in `read_image` tool directly injects image chunks into the current model's context, and it is only usable if the current main model supports images. If your main model is a pure text model such as `deepseek v4 flash`, calls to `read_image` will be rejected—the main model cannot perceive the image, effectively breaking the image reading path. dsh-auto-vision adopts a different approach: it forwards images internally to a multimodal model for recognition, ensuring the main model only processes text throughout the process. This article explains its underlying principles, installation, and configuration. ## What is dsh-auto-vision

dsh-customui-plugin: Add a "Personalization" settings page for DSH Web

# Introduction In the DSH (DeepSeek Harness) web version, several interface elements are hardcoded into the host build artifacts: the brand logo in the top-left corner of the sidebar, the welcome area during empty conversations, and the background color of the conversation area. If you wish to replace them with your own logos or background images in an on-premises deployment, the conventional approach is to modify the host build, which requires redoing the work every time the host is updated. dsh-customui-plugin takes a different approach: it avoids modifying the host build and instead injects scripts into the browser via the plugin mechanism to overlay changes at the DOM layer, while also providing a formal entry point for these modifications in the settings page. The following sections introduce its features, working principles, and installation steps. # What is it? dsh-customui-plugin is a DSH web platform plugin, with the repository address at https://g

dsh-plugin-whale-fenggu: Monitor DeepSeek Peak-Valley Pricing Periods in DSH

# Preface Starting from August 17, 2026, DeepSeek's official V4 pricing follows Beijing time peak and valley rates: expensive during peaks, cheap during valleys. The boundaries at 09:00, 12:00, 14:00, and 18:00 are easy to remember, but the hard part is forgetting to check the time while a long task is running—running heavy work right into peak hours means higher token costs. Keeping track of the time yourself is one way; a more worry-free approach is to let a resident component keep watch for you and alert you before the switch. The dsh-plugin-whale-fenggu introduced in this article is just such a plugin. DSH's philosophy is "everything is a plugin," and these kinds of utility tools fit the plugin system perfectly. # What is it dsh-plugin-whale-fenggu (Blue Whale · DeepSeek Peak

dsh-theme-rheostat: DSH "Rheostat" theme that shifts with model and inference levels

# Introduction DSH's philosophy is "everything is a plugin," and the theme is no exception. However, in daily use, if you switch between deepseek-v4-flash and deepseek-v4-pro or adjust the reasoning level, the interface lacks visual feedback corresponding to the specific setting, making different sessions appear identical. dsh-theme-rheostat, written by nineandnine-9, solves this issue: it maps the current session's model and reasoning level to four distinct visual "gears," ensuring the accent color, glow effects, and the indicator bar above the input box change accordingly. Unlike common static themes, its appearance is not a hardcoded color scheme but is driven by session state; in terms of implementation, it aligns with the maid-atelier theme referenced in the README.

Introduction to DSH Plugins: talon-ui, and how to verify a plugin when documentation is missing.

# Introduction The philosophy of DSH is "everything is a plugin": agent capabilities are mounted via plugins, while plugins are indexed and retrieved through the community directory. Typically, the repository README serves as the basis for writing a plugin introduction. However, for the talon-ui being introduced here, as of the time of writing (September 2026), the README fetch failed, and there is no verifiable plugin content available. Rather than fabricating a functional description, it is better to list the information that can be confirmed, clearly mark what cannot be confirmed, and provide a checklist for verification when data is missing. This checklist is more useful than a conclusion—when encountering a similar plugin in the future, one can simply follow these steps. # What is this The directory entry is named `nsoio/talon-ui`, signed by NSOiO, and the repository address is https://github.co

dsh-plugin-deploy: Deploy to Cloudflare and publish to npm in one line.

# Preface After using an agent like DSH to write a static site, a Worker, or a dsh plugin, things often get stuck at the final mile: to see the online results, you need to register for a cloud account, install Wrangler, write configuration, and configure authentication; to publish the plugin you wrote, you need to check package contents and dependency licenses item by item, and deal with npm's 2FA process. These steps are mechanical, error-prone, and have nothing to do with writing code itself. dsh-plugin-deploy is designed to address this closing work. Below, I will introduce its positioning, installation methods, two main capabilities (deployment and publishing), as well as its design regarding credentials and approvals. # What is it dsh-plugin-deploy is a DeepS... maintained by Octo-o-o-o.

dsh-turn-budget: Adding fail-closed steps, tools, and token budgets to DeepSeek Harness per turn

dsh-turn-budget is a fail-closed resource governance plugin for DeepSeek Harness (DSH), maintained by Nunchakus888 and following the MIT license. It independently sets three types of limits for each Agent turn: model step count, tool execution count (including nested dispatch), and token usage reported by the provider. After the quota is exhausted, subsequent requests are rejected before execution, and the turn is closed with a blocked status. The plugin only uses DSH public extension points and implements interception through agent/pre-step, tools/pre-execute, tool/result, and turn/end hooks, with results preserved in standard Session logs. token 上

dsh-mcp-adapter: Enables DeepSeek Harness to invoke MCP on demand instead of injecting all schemas into the context

# Introduction Connecting MCP servers to DeepSeek Harness (hereinafter referred to as dsh) is most directly achieved via the official `@deepseek-ai/dsh-mcp-client`. This client connects to the server at startup and registers each declared tool as a native `mcp__<server>__<tool>` function. The trade-off is the schema cost: with multiple servers, hundreds of tool definitions are passed into the context with every request, regardless of whether they are utilized. `dsh-mcp-adapter` provides an alternative solution to this problem by registering only a single `mcp` proxy tool. It employs lazy server startup, allowing the model to `search` or `describe` tools on demand.

dsh-keyboard-history: Add ↑/↓ input history for DSH Web inputs

dsh-keyboard-history is a DSH plugin (v0.1.2, MIT License) maintained by NormanFxxkingRockwell that provides terminal-style input history functionality for web session input boxes. Pressing ↑ when the input box is empty recalls the last sent message and allows continued scrolling up, while pressing ↓ goes back.

dsh-skill-router: Rule-based skill routing before step execution

# Introduction When building agents with DeepSeek Harness (DSH), the timing of skill loading is usually left to the model to decide. Skills are exposed as a catalog to the model, visible at every step, forcing the model to re-evaluate whether to load them constantly. The result is unstable decision-making: skills may load too late, load the wrong one, or fail to load entirely. dsh-skill-router shifts this to a rule-driven approach: before each step, pure rules match the latest user message. If a match is found, the corresponding skill content is injected into the current step; otherwise, no intervention occurs. The entire process involves zero LLM calls and zero token costs. Below, we introduce its usage. # What is it? dsh-skill-router is a

dsh-vision-plugin: Add an "Image Input" declaration checkbox to the DSH model directory

# Introduction When integrating custom models into DSH, you may encounter a scenario where the endpoint itself supports multimodal capabilities, but sending an image directly in the chat dialog results in an error stating "Image input not supported." The reason is that the DSH host reads the `input: ["text","image"]` declaration on the model entry to decide whether to pass the original image directly to the model, while the interface in "Settings → Models → Model Catalog" has never exposed an entry point for this declaration. dsh-vision-plugin aims to address this step: by adding an "Image Input" checkbox on the model page in the Model Catalog. Once checked, the declaration is written into DSH settings, allowing the custom model to directly receive images in the chat dialog. # What is it dsh-vision-p

dsh-mcp-sync: Scan local MCP configurations and register tools for direct model invocation.

<tool_call>Bash<arg_key>command</arg_key><arg_value>ls -la "D:\D-Users\D-Administrator\.zcode\workspace\default" | grep -E "obj_|arxiv"</arg_value><arg_key>description</arg_key><arg_value>List obj* and arxiv related directories</arg_value></tool_call>

dsh-dbhub-live: A DBHub bridge plugin enabling DSH to directly and safely operate on databases.

Preface When developing with DSH, allowing the model to query the database directly is a common requirement. There are two typical approaches: pasting the DSN into the conversation for the model to connect itself, which causes the password to enter the model's context; or maintaining a persistent database MCP service yourself, which requires managing ports and process lifecycles. dsh-dbhub-live takes a different path: the password remains only on the host side, while the model is given only a `source` handle. It spawns a one-time process for each query and recycles it immediately after execution. Below, I introduce its design, installation, and usage. ## What is it dsh-dbhub-live is a DSH plugin maintained by mr-mihu, currently version 4.0.0, licensed under the MIT license. It serves as a bridge based on DBHub (the database MCP server):

dsh-llm-ollama: Integrating Ollama Cloud Chat and Web Search Crawling for DeepSeek Harness

dsh-llm-ollama is an MIT-licensed DeepSeek Harness (DSH) integration plugin for Ollama Cloud, maintained by NOirBRight, designed to address the lack of an entry point for editing third-party models on the DSH Models page. Its core features include: enabling Ollama Cloud chat via Ollama's OpenAI-compatible Chat Completions endpoint (via the shared pi-ai adapter); implementing model discovery through native APIs (/api/tags, /api/show) to provide metadata such as context window, vision, thinking, and tools; and registering ollama-cloud's Web Search and Web...

dsh-agent-pill: DSH's agent activity pill and floating summary panel

dsh-agent-pill is a DSH web plugin maintained by mozhuanzuojing (MIT, v0.14.8) that addresses visibility issues during long-running agent tasks: it condenses key session activities into a status capsule in the top-right corner, allowing a floating panel to be expanded via Ctrl+Alt+P or by clicking the capsule. The plugin is divided into a host side (aggregating Goal, Subagents, Agent status, workflow execution, and background tasks, exposing a JSON API via /pill/api) and a client side (the capsule and anchored panel). Main features include a draggable capsule, position persistence, theme following DSH's light/dark mode switching, section folding, and empty state hiding; Goal cards display stages, round progress bars, and workflow history (latest...

dsh-webfile: Adding Approval-Gated S3/FDS/FTP File Tools to DeepSeek Harness

# Introduction The philosophy of DeepSeek Harness (DSH) is "everything is a plugin." Once you need an agent to help you move logs or sync files, remote storage becomes unavoidable: S3, Xiaomi Galaxy FDS, or even still-operational FTP servers. The common approach is to write keys directly into the configuration and let the model call command-line tools within a shell—this exposes keys in the configuration and session logs, and operations like deletion and overwriting are not intercepted. dsh-webfile addresses this issue: it encapsulates three types of storage into eight agent tools. Read operations do not require approval, while change operations request approval one by one. Large file transfers are handled as background tasks that can be canceled at any time. Below is an introduction to this plugin. # What is this? dsh-webfile is DeepSeek H...

dsh-tui: An Interactive Terminal Interface for DeepSeek Harness

`dsh-tui` is a full-screen interactive terminal UI plugin developed by the community for DeepSeek Harness (DSH), serving as a plugin-based supplement...

dsh-plugin-ima-sync: Automatically Upload DSH Conversation Progress to Tencent IMA

dsh-plugin-ima-sync is an MIT-licensed DSH plugin developed by nan1010082085 that automatically uploads DSH Web conversation progress to Tencent IMA's Daily Notes and Work Knowledge Base. It mirrors the local Claude Code `ima-upload` workflow and is designed exclusively for the web platform. Install with `dsh plugin --profile web add`.

dsh-balance: Add a floating model balance widget to the DeepSeek Harness Web GUI

dsh-balance is a DSH Web GUI client plugin designed to persistently display model account balances in the bottom-right corner of the page. It supports four service providers: DeepSeek, SiliconFlow, Moonshot, and OpenRouter, automatically identifying and invoking the corresponding balance interface based on the baseURL domain. Core features include a draggable floating card (displaying total balance, top-up/reward details), automatic refresh every 60 seconds (with a 30-second host-side cache), and a global toggle in the settings page (taking effect immediately). API keys can be provided in two ways: via the settings page or the credentials service. All balance queries are performed on the host side, ensuring keys are never sent to the browser. The plugin has zero runtime dependencies and requires no build process. API routes are protected by loopback and same-origin policies.

dsh-jina: Integrating jina-cli's 12 API tools into DeepSeek Harness

# Introduction DSH's built-in `web_search` covers general web retrieval; however, to obtain arXiv official direct links, SSRN paper originals, or perform web screenshots, text vectorization, and document reordering, one typically needs to write glue code at the host level and register the tools manually. dsh-jina eliminates this step: it encapsulates the Jina AI API capabilities (behind `jina-cli`) into 12 `jina_*` model tools within DSH and provides an API key configuration card with real-time balance checking on the Web settings page. Below, we introduce its features, installation methods, and selection considerations. # What is dsh-jina dsh-jina is maintained by minatoAI, licensed under MIT, and is currently at version 0.5.2.

dsh-peak-cost-mode: Automatic traffic saving plugin for DeepSeek peak-valley pricing periods

dsh-peak-cost-mode is an open-source DSH web plugin (MIT, author moon09300731) designed for DeepSeek V4 peak-valley pricing: peak output prices on weekdays from 9 AM to 12 PM and 2 PM to 6 PM are double those of off-peak hours. During peak hours, the plugin automatically injects a "Peak Cost Mode" prompt segment, instructing the model to respond in a "caveman" ultra-concise style—discarding pleasantries and hedge words while preserving code, commands, paths, and other technical content word-for-word, without compressing safety warnings. In off-peak hours, users can manually enable active cost-saving mode via the savings report, and the switch persists across peak/off-peak cycles and after restarts. The plugin intercepts every streaming call to estimate savings based on actual output tokens and a compression ratio (default 0.6), converting the savings into CNY and categorizing them into peak and active types. Data persists per wave.

dsh-plugin-bilibili: Provides Bilibili search, subtitles, and frame-level viewing capabilities for DeepSeek Harness

# Introduction When integrating information sources for agents, Bilibili is an unavoidable category: finding videos on specific topics, summarizing episodes, and verifying the actual content. The challenge lies in the fact that Bilibili pages are highly dynamic, making direct web scraping unreliable. Providing only the subtitle text to the model is insufficient; when subtitles state "as shown in the figure," the model lacks context on the visual content. The philosophy behind DeepSeek Harness (DSH) is "everything is a plugin," where external capabilities are integrated into the session as tools. dsh-plugin-bilibili implements this approach: upon installation, the agent gains five `bilibili_*` tools covering keyword search, video metadata, subtitle transcripts, direct playback links, and video frames. Crucially, frames are returned as image chunks, enabling models with image capabilities to view the actual visual content directly, rather than relying solely on subtitles.

dsh-liangwengu: DeepSeek API Peak/Idle Time Reminder Plugin (DSH)

# Introduction DeepSeek API pricing is tiered by time zones: 9:00–12:00 and 14:00–18:00 (Beijing Time) are peak hours, while the rest are off-peak hours (half price). It is easy to forget to check the time when busy with tasks. I ran tasks during peak hours all afternoon before realizing it; this kind of thing is not uncommon for API users. Existing solutions involve setting alarms or running cron scripts in the terminal, but they do not follow your working interface. dsh-liangwengu has developed a DSH (DeepSeek Harness) plugin: it pops up a cartoon modal, sends system notifications, and plays sounds to directly tell you "Peak is here, run; Peak is gone, start saving." The following sections introduce its notification mechanism, UI capabilities, and installation method. # What is this

dsh-plugin-model-switcher: Switch models and reasoning levels in DSH using keyboard shortcuts

# Introduction During agent development, frequent comparison between different models is often required: running a task with the current model and then switching models to compare results, or adjusting reasoning levels based on task complexity. Manually opening the model selector, selecting a model, and adjusting levels repeatedly is quite tedious. DSH's philosophy is "everything is a plugin," making high-frequency, small operations ideal for plugins. The `dsh-plugin-model-switcher` introduced below is exactly that: it adds two sets of keyboard shortcuts to the DSH web client—one for switching models and one for switching reasoning levels—accompanied by toast notifications to confirm each action. # What is this `dsh-plugin-model-switcher` by Mrlilil

dsh-model-memory: Supplementing DSH with persistent memory for custom model reasoning levels and channel preferences

dsh-model-memory is a DSH plugin developed by Mutx163, designed to address the issue of missing thinking level configurations for custom API models and the loss of model and thinking strength preferences when switching channels. It mounts a thinking level configuration bar for custom models within the official "Settings -> Models" section, supports low/medium/high/max levels, and allows for hot-reloading of configurations. Additionally, it remembers the last selected model and thinking intensity across sessions per channel, ensuring preferences are preserved when switching channels.

dsh-llm-finish-reason-tolerance: Properly terminating streaming responses missing the finish_reason

# Introduction When connecting OpenAI-compatible models to agents, there is a subtle yet critical compatibility issue: the streaming response from some gateways only pushes content deltas and end markers, without ever sending the `finish_reason` field. DSH's model client, pi-ai, classifies this stream as truncated ("Stream ended without finish_reason"), which the host then maps to a `TRANSPORT` error finish. The result is that while the content of every request is delivered intact, the turn ends in failure. A typical example is the Snowflake Cortex REST gateway (`*.snowflakecomputing.com/api/`)

dsh-patchouli: Local Memory and Knowledge Hub for DeepSeek Harness

# Introduction When developing agents with DeepSeek Harness (hereinafter referred to as DSH), the "everything is a plugin" philosophy implies many choices for memory capabilities. Hindsight, MemOS, and Graph Memory, listed in the compatibility table, all belong to this category. The problem is that each memory plugin comes with its own storage and retrieval algorithms, and their integration methods are different. To use them together in the same environment, one must adapt interfaces individually and maintain data separately; switching implementations requires modifying the integration layer. `dsh-patchouli` addresses precisely this issue: it decouples "data integration" from "algorithm implementation" by placing a unified memory and knowledge service in the middle. Below is an introduction to this plugin. # What is this? The positioning of memorax-ai/dsh-patchouli is Deep...

dsh-blue-archive-shiroko: Blue Archive Theme and Shiroko Desktop Pet for DSH Web Client

`dsh-blue-archive-shiroko` is an unofficial fan-made theme plugin for the DSH Web Client developed by mldhao (v0.6.0, MIT License), providing a Blue Archive style visual and interactive experience for DeepSeek Harness. Core features include: a complete UI theme centered around academy blue, cyan highlights, and white frosted glass panels; a Shiroko desk pet that supports dragging, resizing, and autonomous interaction; petting interaction and sound effects triggered by holding the head for approximately 420ms; and Codex-style short bubbles with streaming character-by-character text display. In terms of utility, a short press on the desk pet allows querying the API balance via a Host-side RPC call to the DeepSeek official interface. The API key does not enter the browser, and the RPC is restricted to loopback address calls, ensuring security. Installation requires the DSH plugin system.

dsh-unsloth-hands: Offloading repetitive tasks of DeepSeek Harness online models to local Unsloth Desktop

dsh-unsloth-hands is a pure client-side plugin maintained by MicroHEROX (MIT licensed, v0.1.0), allowing DeepSeek Harness online models to perform repetitive short tasks such as batch rewriting, translation, string cleaning, and screenshot OCR.

catalog-capabilities-zh: Maintain a source-based Chinese capabilities catalog for DSH

catalog-capabilities-zh is an MIT-licensed community project maintained by MeowTnT3r. Acting as a DSH bundle plugin and runtime skill, it maintains a source-based Chinese capabilities catalog for skills, plugins, and marketplace capabilities. It follows the "everything is a plugin" philosophy, addressing the pain points of scattered DSH user plugins, predominantly English READMEs, and difficulty in finding them. Its core feature is a default curated mode where only user-confirmed entries are written to the documentation; scanning serves only as a discovery aid, and explicit full inclusion requires the `--all` or `full` mode. It does not rewrite installation logic; instead, it orchestrates existing trusted installers in the environment during installation. If the target is already installed, it only verifies update instructions. When only providing a link, it reads public repository metadata for checks and introduction without executing the repository code. Entries support sorting by Chinese...

dsh-vulnsec-bridge: A security plugin for integrating VulnClaw and DeepSec into DeepSeek Harness

# Introduction When performing security-related tasks in DeepSeek Harness (dsh), the common practice is to open a separate terminal to run penetration testing or code auditing CLIs, then copy the output back to the session for further analysis. As tool capabilities are not registered as dsh tools, the model cannot invoke them directly, and their results cannot be integrated into subsequent automated workflows. dsh-vulnsec-bridge is a bridging plugin designed for this scenario: it wraps VulnClaw (AI penetration testing CLI) and DeepSec (Shield code auditing + Spear authorized penetration) into dsh plugins, providing 18 security tools and a graphical "Security Workbench" panel. Below, we introduce its features, installation methods, and typical usage. First, let's discuss the prerequisites: the plugin's penetration...

dsh-terminal: A DSH terminal plugin awaiting verification

# Introduction DSH (DeepSeek Harness) operates on the philosophy of "everything is a plugin": capabilities such as terminal access, browsing, and file operations can be augmented into agents via third-party plugins. A member of the community maintains a plugin directory that catalogues numerous such plugins, and dsh-terminal is one of them. However, this introduction will be brief. The reason is straightforward: as of the writing of this article (2026-09-07), the scraping of this plugin's GitHub README failed, and I lack first-hand materials to verify its functionality, usage, and license. Rather than fabricating a detailed feature list, I have placed the confirmed information and links here, leaving the rest for you to verify in the repository yourself. # Confirmed Information The following points are taken from the entry in the plugin directory and are confirmed to be valid:

dsh-session-stars: Cross-Workspace Star and Reopen DeepSeek Harness Sessions

dsh-session-stars is a Web-only plugin for DeepSeek Harness Web, maintained by malevrigns and licensed under the MIT license. It addresses the pain point where useful Sessions are scattered across different Workspaces, making them difficult to find again. The plugin provides a global cross-Workspace favorites center. Users can favorite a Session with one click in the Session title bar, access the favorites center from the bottom of the sidebar or the collapsed track, browse all favorites with a priority on the newest, reopen still-active Sessions, and explicitly remove expired entries. The title, Workspace, running, and archived status of favorite entries are rendered from real-time DSH snapshots, and synchronization across multiple tabs is automatic via the `storage` event. Installation requires placing the plugin...

dynamic-schulte-dsh: A Schulte Grid plugin for the dsh Web UI to play while waiting for the model

`dynamic-schulte-dsh` is a waiting-time mini-game plugin for DSH, developed by marvin9551. It allows users to play the Schulte Grid within the Web UI while waiting for the model to respond during a DeepSeek Harness intelligent conversation, making use of fragmented time ranging from a few seconds to dozens of seconds. Adhering to the DSH philosophy of "Everything is a plugin," it is installed via `dsh plugin add`, registers a "Schulte" entry at the bottom of the Web UI sidebar, and features a breathing green dot indicator when the model is thinking. Core features include a classic n×n static grid (3×3 to 7×7), a rotating dynamic roulette wheel, waiting prompts, localStorage-based local persistence for history and settings, and optional number highlighting and color distraction enhancements. The architecture is split into two halves: the host part is a purely placeholder plugin.

DSH Claude Style Reasoning Slider: Give DSH a New Animated Reasoning Level Slider and Model Selector

# Introduction In DSH (DeepSeek Harness), switching models and reasoning intensity relies on the native model selector within the session input area. In practice, there are two issues: first, different vendors have inconsistent naming conventions for levels, requiring users to memorize mappings for terms like `off`, `med`, `extreme`, and `ultracode`; second, custom routers (e.g., OpenRouter) often lack `reasoningEfforts` configurations, leaving the interface without an entry point for adjusting levels. DSH's philosophy is "everything is a plugin," meaning that the model slot in the session input can also be taken over by a plugin. The DSH Claude Style Reasoning Slider introduced in this article does exactly that:

@max-null/dsh-chat-rail: Canvas-style message navigation bar for DSH sessions

`@max-null/dsh-chat-rail` (v0.6.1, MIT) is a DSH plugin maintained by Max-Null, positioned as a "scrolling-style message navigation bar" to address the issue of repeatedly scrolling through long sessions to find historical messages. It places a vertical navigation bar on the right side of the conversation panel, featuring an indicator dot for each user message. It smoothly expands the full list on hover, and clicking allows jumping to the specific message. Key features include: `scroll-spy` to highlight the reading position and automatically center it; synchronization with `dsh-better-sidebar` via CSS variables to avoid overlap; deep history navigation using the official precise loader on kernels 0.1.2-alpha.3+, with an automatic fallback to page-by-page loading on older kernels; a tooltip preview panel that pops up when hovering over a message item; and support for starring/favoriting (stored in localStorage).

dsh-opg-usage: View OpenCode Go plan quota in real-time within the DSH Interface

## Introduction When running tasks using the OpenCode Go plan, checking remaining quota and reset times typically requires leaving the current interface to look elsewhere. Checking once is insufficient; you also need to remember the time of your last check to estimate consumption, or you risk accidentally hitting the limit. dsh-opg-usage solves this problem by integrating the quota panel directly into the DSH interface and maintaining a persistent floating capsule for instant checks. Below, we introduce its features, installation, and usage. ## What is it? dsh-opg-usage is a DSH Web plugin maintained by M4sh1r0-444 and licensed under the MIT License. It is a draggable, collapsible floating capsule that displays the usage percentage of the OpenCode Go plan across three quota windows: rolling window, this week, and this month.

june-agent: June Dual-Personality Agent and Supporting Skills, installable as a DSH plugin

june-agent is an npm package maintained by lemonorangeapple that packages two fixed-personality Agents and two skills into a cross-hosting solution. June-M features a "catgirl" personality (temperature 0.1, obedient and concise), while June-S features a "dominant" personality (temperature 0.3, disciplinary pushback). The skills include Git backup strategies (git-backup) and design coding standards (june-style). The package provides both Claude Code format (agents/ and skills/ directories) and DeepSeek Harness format (presets/ directory), DSH.

dsh-office-cli: Securely expose official WeCom, DingTalk, and Feishu office CLIs using a DSH tool.

## Introduction If you have built agents on DeepSeek Harness (DSH), you have likely encountered requirements such as letting the agent send a WeCom message, check a Feishu document, or create a DingTalk approval. The official CLIs possess these capabilities, but allowing the model to execute commands directly in the shell exposes authentication commands, credential parameters, and arbitrary paths to the model, making it difficult to control permission boundaries. Below, I introduce dsh-office-cli, a plugin that safely exposes the three official office CLIs (WeCom, DingTalk, and Feishu/Lark) through a single native DSH tool. ## What is it? dsh-office-cli is maintained by meliwanx and is released under the MIT license. When...

dsh-tool-vision-read: A DSH plugin that enables text-only agents to "see" images

# Preface In agent development, a common gap arises: the model router for the main agent does not declare image inputs and is a pure text model. When users provide an image path for description, the agent cannot handle it—requiring manual switching to a session with vision capabilities or the agent manually viewing the image and relaying the information. The philosophy behind DeepSeek Harness (hereinafter referred to as DSH) is "everything is a plugin." This specific capability gap is ideally suited to be patched by a small plugin. Below, we introduce `dsh-tool-vision-read`, maintained by Mappedinfo. It registers a `vision_read` tool to route the image reading step to a dedicated vision model, enabling a pure text agent to obtain a textual description of the image with a single call.

dsh-subagent-setting: Unified configuration of subagent default model in web settings page

## Introduction If you run multi-subagent task flows in DeepSeek Harness (DSH), you might encounter a specific issue: the default configurations for a subagent's provider, model, and reasoning effort are scattered across configuration files. To switch to a different default model, you must manually edit the YAML file and then verify that every subagent has adopted the new values. `dsh-subagent-setting` brings this to a web settings page: select the default provider, model, and reasoning effort on the interface, and newly created subagents will take effect immediately. You can also choose to apply this to existing subagents.

dsh-deepseek-status: Display balance and peak and valley pricing in the official DeepSeek session header

# Introduction When using DSH to run official DeepSeek sessions, there are two things I need to repeatedly switch away to check: first, how much API balance remains, which requires opening the platform usage page elsewhere; and second, whether I am currently in the peak billing period. DeepSeek API uses peak-valley pricing, where peak rates (full price) and off-peak rates (half price) differ, so it is best to know this before scheduling batch tasks. Neither task is complicated, but switching away from the session to check each time interrupts the workflow. `dsh-deepseek-status` directly embeds these two statuses into the session header: balance, recharge link, and peak-valley pricing status—three capsules that are immediately apparent. Below, I will introduce its features, installation method, and a few usage notes. # What is this? `dsh-deepseek-status` is a DeepSeek status tool.

dsh-plugin-context-manager: A DSH plugin that automatically records conversations, injects memory, and folds real context.

# Introduction Running long sessions in DeepSeek Harness (DSH) often presents two common challenges: First, the limited context window means that once early conclusions are squeezed out, the model "forgets" them; Second, when switching to a different session, previously recorded key points cannot be carried over. The common solution is to inject summaries into the system prompt, but this only affects auxiliary prompt blocks, while the real dialogue history continues to grow, leading to increasing token consumption. The `dsh-plugin-context-manager` introduced below addresses these two issues: it automatically records every round of dialogue, injects memory and prioritized cross-session notes into each model request, and directly collapses the real dialogue history visible to the model to reduce token usage. # What is dsh-plugin-context-manager

lwklbb/dsh-worktrees: Integrate Git worktree management into the DSH settings panel

# Introduction When developing with DSH, you often need to handle multiple branches of the same project simultaneously: keeping the main line on `main`, opening a branch for features, and another for fixing production issues. While Git worktrees solve this pain point by checking out multiple branches of the same repository into different directories, using the command line requires remembering paths and arguments, and managing scattered paths across the disk can be cumbersome. The `dsh-worktrees` introduced below brings this functionality into DSH's settings panel. # What is this? `dsh-worktrees` is a DSH plugin maintained by lwklbb, released under the MIT license, with the current version being 0.1.0. Its positioning can be summarized in one sentence: centralized management of Git worktrees in the DSH settings panel.

dsh-cinematic-workflow: Authorial Cinematic Workflow Plugin (Manual Gating)

# Introduction Running media tasks in DSH (DeepSeek Harness) often leads to stage loss of control: if you instruct the model to perform a rough cut of a travel film, it might process the analysis, editing, color grading, and final cut all in one go, without any intermediate checkpoints for your confirmation. Relying on a prompt like "ask me before each step" is unreliable in multi-stage, long workflows. `dsh-cinematic-workflow` is designed to address this specific issue. It implements a cinema-grade travel film director workflow as a DSH plugin and embeds manual gating directly into the code: the only entry point for advancing stages is the human. Below is an introduction to its positioning, workflow structure, and usage. # What is it `dsh-cinematic-workflow` is a tool oriented towards DeepSeek...

dsh-multi-root: Mounting VS Code-style multi-root workspaces for DSH Web GUI

# Introduction When developing agents using the DSH Web GUI, a common frustration is that there is only one workspace: code resides in one directory, datasets in another, and references in a third. To enable the agent to operate across these directories, you must either manually move files into the same workspace or simply prevent the agent from accessing them. `dsh-multi-root` addresses this problem. Below is an introduction to the plugin: it adds a `Roots` (multi-root) panel to the sidebar, allowing an arbitrary number of independent folders to be mounted as workspace roots. Then, through a set of controlled host-side tools, it enables the agent to list/read/write/glob across these roots—drawing inspiration from VS Code.

dsh-automode: Adding an auto mode with deterministic guardrails to DeepSeek Harness

To enable an agent to execute tasks continuously, there are typically only two permission options available: using a conservative preset that halts for manual confirmation at every write operation, or switching directly to danger-full-access. While the latter is faster, it grants the agent unrestricted autonomy without any safeguards if issues arise. DeepSeek Harness (DSH) permission presets—read-only, workspace-write, and danger-full-access—exist at these two extremes, lacking a middle tier that is "loose but has gatekeepers for dangerous actions." The dsh-automode introduced below fills this gap: it is a DSH plugin that provides a Claude Code-style auto mode, allowing...

deepseeh-harness-ultra-slash: A separate group of plugin commands in the DSH / menu

# Introduction People using DeepSeek Harness (hereinafter referred to as DSH) largely rely on the `/` menu in the chat input box. DSH's philosophy is "Everything is a plugin." When many plugins are installed, finding commands becomes a practical issue. Another high-frequency need occurs mid-conversation: you think of a new instruction you want the model to process in the next step but do not want to interrupt the ongoing dialogue. The loadingvx Ultra Slash plugin handles both of these simultaneously: it separates plugin commands into a separate group in the `/` menu, distinct from DSH's built-in commands, and provides four commands: `/steer`, `/new`, `/skill`, and `/docs`, as well as the ability to assign short names to frequently used `/steer` content. The following introduces its features, installation, and...

dsh_PromptRecall: Recall historical prompts in DSH Web input box using ↑/↓

dsh_PromptRecall is a DSH Web GUI input history plugin that mimics Codex interaction. Maintained by liguanyu (MIT license, version 1.0.0), it addresses the lack of terminal-like history functionality in web input fields. Once installed, it requires no configuration; pressing `↑`/`↓` while the input field is empty allows browsing historical prompts, which are persisted across sessions and restarts. If the draft is non-empty, the cursor is in the middle of the text, a selection is active, or a modal is open, the direction keys are returned to the editor. Pressing Esc clears the draft and saves it to history, allowing for later retrieval. The history is designed in two layers: the persistent layer is a JSONL file located in the DSH main directory (with a limit of approximately 10 MB for plain text only), and the local layer is the recall list for the current session (with a limit of 200 entries or 256 KB).

dsh-workbench-layout: Provides a three-column workspace layout for DSH Web

dsh-workbench-layout is a DSH Web plugin (npm package @lsq64737/dsh-workbench-layout, MIT license) that reorganizes the official AppFrame into a three-column layout: left side contains Sessions, File Browser, Git Changes and Commit Graph, Terminal List; middle is Multi-file Tab Editor and Terminal; right side preserves the native DSH session, allowing users to view files, Git status, and run commands without leaving the conversation. Core features include: sorting by Git status...

dsh-cache-hit-precision: Display DSH status bar cache hit rate with two decimal places.

# Introduction Under the status bar in dsh web, there is a line of statistics, and the cache hit rate is one of them. By default, this value is not displayed with two decimal places, which makes it difficult to observe subtle changes between consecutive rounds. To modify such displays, directly editing the code in the dsh installation directory is one approach, but it mixes your changes with the upstream environment, making maintenance difficult. The philosophy of DSH is "everything is a plugin," and these kinds of minor interface changes are perfect candidates for a plugin solution. The following introduces just such a small plugin: dsh-cache-hit-precision, created by luern0313. # Plugin Introduction dsh-cache-hit-precision targets the dsh web platform (the `dsh. property in package.json.

dsh-plugin-forge: Manually-gated Workbench for DSH Plugin Creation, Validation, and Publishing

dsh-plugin-forge is a DSH plugin (v0.1.0, MIT) maintained by luoyuejun9, designed as a human-gated workspace for plugin creation, verification, and release. It weaves error-prone stages in plugin development—scaffolding, type checking, testing, packaging and previewing, key scanning, clean environment installation verification, and publishing—into a replayable workflow. It offers four templates: tool-command, skill-wrapper, stateful, and bundle, and is operated via the /forge command and Web Studio cards within a chat interface. Card states are reconstructed based on persistent session events and persist even after the conversation is replayed. The core security design ensures that stage progression relies solely on human confirmation; only /forge continue can advance a stage, while the model can read the state and perform validation,

dsh-token-monitor: DSH Web Large Model Quota and Usage Monitor Plugin

It is standard for DSH users to connect to multiple model providers simultaneously: Kimi For Coding utilizes subscription quotas, DeepSeek Official deducts balance based on usage, and OpenRouter tracks credits. To check "how much is remaining" or "how much has been spent this month," users typically have to log in to each provider's console separately, toggling back and forth. By the time the round of checks is complete, work at hand is often interrupted. dsh-token-monitor integrates this functionality directly into DSH Web: it features a persistent balance badge in the session header, an additional "Usage" tab in the main area, and local SQLite records tracking token usage and estimated costs for each API call. Below, we introduce its features, installation, and configuration. # What is this? dsh-token-monitor is a component of DeepSeek Harness (

dsh-greet-plugin: A minimal greet tool plugin for DeepSeek Harness

dsh-greet-plugin is a minimal plugin example of DeepSeek Harness (dsh) maintained by luxiu666 (MIT licensed). It adds a greet tool to an agent that accepts a `name` argument and returns a greeting message. The default greeting "Hello" is stored in `cordis.patch.yml` and can be overridden via the profile patch layer without modifying the code. The repository includes pre-built artifacts and a build-free installation script. The command is `dsh plugin --profile web add github:<you>/dsh-greet-plugin#v0.1.0` (<you> is a placeholder). Requires Node.js ^22.19.

dsh-cc-switch: One-click sync cc-switch transit merchant configurations to DSH model routing

# Introduction Running agents with DeepSeek Harness (DSH) inevitably involves model configuration. If you have been using cc-switch to manage proxy configurations for Claude Code / Codex, these ready-made providers cannot be used directly within DSH. The conventional approach is to manually re-enter API addresses and keys in DSH and set up the routes. dsh-cc-switch is designed to solve this: it enables a one-click synchronization of the provider configurations managed in cc-switch into DSH's model routes, without requiring any models to be pre-configured in DSH. In other words, a brand new DSH instance with zero models can get started using your existing providers via this tool. DSH's philosophy is "...

oh-my-dsh: DeepSeek Harness Layered Model Routing Plugin for think/build

# Introduction There is a common dilemma when running agent coding sessions: running the strongest model at every step is effective but costly and slow, whereas using cheap models throughout cannot handle complex tasks like planning and architecture design. Manually switching models between sessions is equally cumbersome and interrupts the flow of thought. oh-my-dsh solves this problem: it implements model tiering within a single DeepSeek Harness (DSH) session, using strong models for planning and cheaper models for daily execution, with dedicated channels for image understanding and image generation. This eliminates the need for manual session switching or YAML editing. The following sections introduce its features, installation methods, and typical usage. # What is oh-my-dsh? It is a DSH plugin maintained by llmpolska under the MIT License. The current version is...

dsh-hero-custom: Customize DSH new conversation homepage title in settings page.

dsh-hero-custom is a DSH (DeepSeek Harness) Web GUI plugin designed to customize the "Explore the Unseen: Preview" title area of the new chat home page without modifying official frontend files. Maintained by Lumera114514 and released under the MIT license, it allows users to modify main title text, logo, font size, color, font, position, neon glow effects, title effects (flowing light, breathing, rainbow), as well as the size, color, and animation of the whale icon in the "Settings -> Main Interface Title" page. The plugin supports real-time preview and auto-save. Its anti-flicker mechanism prevents old text from appearing briefly. Settings are persisted locally to `$DSH_HOME/hero-custom.settings.json`, survive restarts, and do not upload any data. Installation requires Node >= 20 via npm.

dsh-molbio-tools: Zero-dependency Molecular Biology Research Plugin for DeepSeek Harness

dsh-molbio-tools is a zero-dependency molecular biology plugin for DeepSeek Harness (DSH) containing 44 pure .mjs tools covering sequence analysis, automated PCR primer design and checking, restriction enzyme digestion and cloning assembly simulation (including Golden Gate), GenBank/SnapGene parsing and SVG plasmid maps, Sanger sequencing verification, qPCR analysis, codon optimization, multiple sequence alignment and conservation analysis, and PubMed.

dsh-lan-proxy: A reverse proxy plugin to expose the dsh Web UI to the LAN

# Introduction By default, the Web UI of DSH listens only on the loopback address, meaning it is only accessible from the local machine. To open it on another computer or tablet on the local network, one typically needs to set up a reverse proxy: starting a separate process, modifying configuration files, handling token links and WebSocket forwarding, and dealing with the hassle every time DSH restarts. dsh-lan-proxy turns this into a plugin: switches, status, and startup logs are embedded directly into the DSH settings page. It starts and stops with DSH, eliminating the need to maintain a separate process. Below is an introduction to its features, installation, and usage. # What is it? dsh-lan-proxy is a DSH plugin maintained by liveqte and is released under the MIT license. Its core function is to change the listening address from 127...

dsh-api-balance: View DeepSeek Balance and Real Token Usage in the DSH Settings Panel

dsh-api-balance is a DSH plugin developed by gaozhiwei521 (v0.1.0, MIT) that addresses the lack of a public usage interface for DeepSeek, requiring developers to log in to the console or manually estimate tokens.

dsh-toolkit: One command to equip DSH with MCP management, Skill management, and Git branch badges

# Introduction If you are using the DeepSeek Harness (DSH) web client, you are likely familiar with the following operations: configuring MCP servers requires manually editing `$DSH_HOME/profiles/web/cordis.patch.yml`; adding or removing Skills requires directly modifying Markdown files under `$DSH_HOME/skills`; and checking which git branch the current session is running on requires switching back to the terminal. These tasks are not difficult in themselves, but because they are scattered outside the client, the overhead of switching back and forth is significant. dsh-toolkit aims to solve this problem by packaging three functions—MCP management, Skill management, and Git branch badges—into a plugin toolkit.

dsh-api: An HTTP control plane layer on dsh's listening ports

When using scripts, desktop wrappers, or editor plugins to control `dsh`, a common issue arises: capabilities such as changing languages, switching workspaces, and detecting when an agent is idle are embedded within the `dsh` process itself. External processes either directly access its in-process services or create their own communication channels, leading to inconsistent interfaces and high coupling. The `dsh-api` approach consolidates this into a single entry point: since `dsh` already listens for HTTP on 127.0.0.1, plugins can directly mount a set of `/dsh-api/*` JSON routes on this socket. Any process on the same machine—desktop wrappers, browser extensions, CLI tools, or editor integrations—interacts through this channel. The following sections introduce its positioning, interfaces, installation, and common usage scenarios.

dsh-trace-insight: A read-only review plugin that organizes DSH execution traces into a traceable analysis timeline

# Preface When executing long tasks with DeepSeek Harness (DSH), native traces capture messages, tool calls, and events, but a single task round often involves hundreds or thousands of steps. It is not easy to answer a few basic questions: What strategy did the agent actually adopt? Which steps are progressing the task, and which are repeated attempts? Do failures stem from model judgment, tool usage, or environmental conditions? Does a "completed" answer have sufficient evidence? Scrolling through traces one by one is inefficient. What dsh-trace-insight introduced below does: it organizes dense trace events into a continuously updated analytical timeline, ensuring that every conclusion can be traced back to the corresponding original event. # What is it dsh-trace-insight (DSH Trace Insig

dsh-nba-pets: Use basketball pets to view task status in DeepSeek Harness

# Introduction When running tasks with DeepSeek Harness (hereinafter referred to as DSH), a common trouble is that the task executes in the background, and you have to switch windows or refresh the page to know if it is waiting for input, has encountered an error, or has finished. `dsh-nba-pets` solves this problem in a different way: it places a basketball-themed desktop pet on the DSH page that uses different animations to directly express the task status. When input is required, it waits for you; when an error occurs, it displays a blocked action; and when idle, it is there practicing shooting. Below is an introduction to the plugin's positioning, functions, and installation methods. # What is it? `dsh-nba-pets` is a collection of basketball-themed desktop pets maintained by lijian-888, built specifically for DSH. It is licensed under the MIT license, and the repository is `packa

dsh-3d-model-viewer: View 3D Models in DSH, enabling LLMs to understand parts.

# Introduction When using Large Language Models (LLMs) to address mechanical and structural problems, a common obstacle is the model files themselves. Formats such as OBJ, STL, and STEP are either pure geometric data or BREP entities that require a CAD kernel for discretization, meaning pure LLMs often receive nothing but a stream of bytes they cannot understand. Previous solutions either required running a local CAD toolchain to extract results or simply gave up. `dsh-3d-model-viewer` is a plugin in the DeepSeek Harness (DSH) ecosystem. The approach involves splitting the process into two steps: first, rendering the model directly in a web interface for human visualization, and second, translating the same model into a standardized, LLM-readable JSON to be passed to the language model. The following sections introduce its positioning and installation.

dsh-deepseek-balance: Real-time view of DeepSeek API balance and usage fees at the bottom right of DSH

# Introduction When running agent tasks with DSH (DeepSeek Harness), checking API balances and costs typically requires opening the official console separately: switching tabs, logging in, and scrolling through invoices disrupts your workflow. To conveniently keep track of your "remaining balance" and "daily spending," you need a monitor that persists within the client. The dsh-deepseek-balance plugin introduced below solves this problem: it places a floating badge in the bottom-right corner of the window, displaying the account balance in real-time along with usage and cost charts for the past 7 and 30 days. # What is it dsh-deepseek-balance is a DSH plugin maintained by lin-cheng-lab, currently at version 0.1.

dsh-ptc-cordis-preset: Add Creative Mode to DSH PTC

# Introduction When developing with DSH (DeepSeek Harness), there are two types of tasks that are very common. One type is tool orchestration: using the Code Mode SDK to write a `run_code` program to combine multiple tool calls into a single execution. The other type is creative work centered around the runtime: inspecting the live runtime, defining and experimenting with dynamic plugin packages, and even creating new agent presets. These two capabilities currently belong to two built-in presets in DSH: the built-in PTC preset (ID is `code`, renamed to `ptc` starting from dsh 0.1.2) provides Code Mode orchestration; the built-in creative preset (`cordis`) provides self-referential Cordis tools...

dsh-web-speech-input: Add a microphone button to the DeepSeek Harness Web input field

dsh-web-speech-input is a speech-to-text plugin for the DeepSeek Harness Web UI (maintained by leozou320-ai, v1.0.0, MIT license). Its core functionality is adding a microphone button to the left side of the prompt composer. Upon clicking, it uses the browser's Web Speech API to write real-time speech recognition results into the current draft. It supports interim and final recognition text, automatic language selection (Page Language -> Browser Language -> zh-CN fallback), clear status indicators, and accessibility support. The plugin never sends a prompt automatically; the submission action is always performed by the user. In terms of privacy, microphone permission is only requested after the user clicks. There are no server-side components, no audio storage, and no network requests are sent; audio processing depends on browser implementation. Installation Requirements

dsh-taffy-theme (Taffy Live Studio): Apply a Pink and Gold Acrylic Theme for DSH Web

# Introduction When using DSH Web for daily development, if you want the interface to be more distinctive—for example, for live streams, screen recordings, or simply to change the mood—modifying the frontend styles directly is quite costly. DSH's philosophy is "everything is a plugin," and styling can be handled by plugins as well. The `lengzhanbao/dsh-taffy-theme` introduced below is a theme plugin designed solely for changing the look: install it to change the skin, and uninstall it to revert to the original. # What is this? `dsh-taffy-theme` (full name: Taffy Live Atelier / 塔菲直播工房), created by lengzhanbao, is categorized under "Fun Dress-up" in the community directory. It provides a virtual livestream workshop-style interface for DSH Web: a light "Flower House" and a dark...

dsh-lark-bridge: Get notified on Feishu when DSH stops working

**Introduction:** Running a long DSH session is most troublesome not because tasks are slow, but because it tends to silently halt and wait for you: the model may be stuck on `ask_user_question` awaiting a response, tools may be pending sandbox upgrade approvals, or rounds may end abruptly due to 4xx errors. While you are occupied with other matters, you might not notice until returning ten minutes later that it has already stopped. `dsh-lark-bridge` addresses this issue by invoking `lark-cli` in real-time to send Feishu/Lark notifications whenever a DSH session ceases working. The goal is to ensure that "DSH stops working = Must receive notification." Unlike existing methods (screen polling or scripts covering only single events), it covers scenarios such as waiting for interaction, task completion, being blocked, request backoff, process stagnation, and process termination.

dsh-mcp-lazy: DSH's Lazy-Loading MCP Plugin, Reducing Token Waste by Disclosing Tool Schemas on Demand

# Introduction The philosophy of DSH (DeepSeek Harness) is "everything is a plugin." Developers often install multiple MCPs on it, such as files, browsers, and databases. This incurs a specific cost: even if the current problem only requires file tools, all MCP tool names, descriptions, and parameters may enter the model context simultaneously. The more MCPs installed, the more tokens are consumed per round. Below is an introduction to leaforbook/dsh-mcp-lazy, a DSH plugin designed to hide descriptions of temporarily unused tools and load them when the task requires them. # What is it dsh-mcp-lazy (npm package name `@yilinxiao/dsh-mcp-lazy`) is an MCP for DeepSeek Harness (DSH).

dsh-task-complete-notifier: Pops up a floating card in the bottom right corner that allows command continuation upon task completion.

# Introduction Running agent tasks with DSH often takes anywhere from a few minutes to tens of minutes, especially during multi-round goal loops. During this time, you likely switch to doing other things. To know when the task ends, you either have to occasionally switch back to the window to check, or write a polling script yourself—the former wastes attention, while the latter is a roundabout way of doing things. `dsh-task-complete-notifier` solves exactly this problem: when a task actually finishes, it pops up a pinned card in the bottom right corner of the screen to notify you, and you can even directly type a command on the card. DSH's philosophy is "everything is a plugin," so this kind of client-side notification requirement fits perfectly as a plugin. Below, I introduce its features, installation, and usage. # What is it dsh-task-complete-notifier is a project by Krea.

dsh-fleet-audit: A read-only credential hygiene audit for DSH agent machines

## Introduction In the era of multi-agent development environments, credentials are scattered across `~/.gitconfig`, agent configurations, `.git/config`, and various `.env` files. If Git's `url.*.insteadof` or `pushurl` settings embed a token (e.g., `https://oauth2:<token>@gitlab.example.com/...`), every invocation of `git remote -v` will expose the secret in logs, chat sessions, or CI pipelines. The traditional approach involves manually checking file permissions and grepping remote configurations one by one, which is time-consuming and error-prone. dsh-fleet-audit automates this process...

@kongxiangyiren/dhs-theme-plugin: Theme Management Plugin for DSH Web Interface