全部插件 · 第 80 页

@qi-seven33 Silver ? Unverified

Optional JEV decision mode for DeepSeek Harness Desktop: deterministic reads, bounded semantic selection, native authorization and credential UI.

@sculptor815 Silver ? Unverified

One plugin, six agent skills for guided single-cell RNA-seq analysis in Codex, Claude Code, and DeepSeek Harness. And one benchmark for evaluating agent performance on scRNA-seq data analysis .

@yellowbee686 Silver ? Unverified

Composable context and recursive language model plugins for DeepSeek Harness

@yum-wu Silver ? Unverified

Pi-style Code Mode for DeepSeek Harness (DSH): Programmatic Tool Calling via QuickJS-WASM sandbox

@qlheric Silver ? Unverified

Runtime supervision for DeepSeek Harness agents: 4 cognitive guards (task contract / loop detection / evidence-before-done / single injection) + 1 cost guard. No core changes. Measured: cuts stalls -48% at only +2% cost.

@hy-sde Silver ? Unverified

Agent Graph for DeepSeek Harness (durable supervisor): control plane, operator executor, stream layer, projection, wakes, host assembly and supervisor tools — the dsh-graph-* / dsh-tool-graph packages.

@piecuzwhynot Gold ? Unverified

Give your DeepSeek Harness agent three more pairs of hands. It sends long tasks to Claude Code, Codex and Antigravity using your own subscription logins, keeps one persistent session per lane, and never loses a job. 给 DeepSeek Harness 用的外部 agent 泳道：把长任务派给 Claude Code、Codex、Antigravity 三个 CLI，用你自己已有的订阅登录。每条泳道一个持久会话，作业重启和压缩都不会丢。

@hy-sde Silver ? Unverified

Automatic long-term-memory extraction at compaction checkpoints for DeepSeek Harness: evidence projection, proposal/canonicalization pipeline, gated ctx.memory writes.

@kyan001 Gold ? Unverified

Ask TypeSafe's Jev how deep a prompt needs to think. Apply that level to the model request.

@yytbit Silver ? Unverified

Jev-scored context compaction for DeepSeek Harness. Score every message by relevance to the current task instead of dropping the oldest, pin the tracebacks and constraints, and fall back to stock behavior on any failure.

@yytbit Silver ? Unverified

Per-turn model and reasoning-effort routing for DeepSeek Harness, decided by one Jev choice question. Jev picks the candidate, a confidence gate falls back, and every turn appends a receipt that can be replayed at zero cost.

@ymrwithnoworry Gold ? Unverified

A general-purpose prompt optimizer plugin for DeepSeek Harness: turn a casual composer draft into a clear, structured, executable prompt — reviewed by the user, written back to the draft, and never sent automatically.

@xeraph627 Silver ? Unverified

为短上下文窗口的对话提供更轻量的 Agent 预设，能降低78%的固定开销。由 Deepseek 制作

@alive114514awa Silver ? Unverified

Guided prompt optimization for DeepSeek Harness: Harness AI and Codex, editable previews, and explicit draft insertion across desktop, Web and TUI.

@yuanstarstar Gold ? Unverified

DSH 女仆模式插件

@lihanhong2002 Gold ? Unverified

MiniMax multimodal tools for DeepSeek Harness: image understanding, image generation, speech synthesis, and transcription.

@clnum42 Gold ? Unverified

在 DSH 原生面板里聊 DeepSeek 网页端，并一键迁移成 harness 会话

@montersy123 Gold ? Unverified

Browse, search and install SkillHub skills inside DeepSeek Harness. / 在 DeepSeek Harness 内浏览、搜索并安装 SkillHub 技能。

@himi-li Gold ? Unverified

Detects verbatim reasoning loops in DSH LLM streams and cuts them early. Zero-dependency Cordis plugin with a fire journal and a reasoning_loop_log tool.

@stefanohe Silver ? Unverified

DeepSeek Harness DataSecure Remastered is a data-security-focused remastered edition of the official open-source DeepSeek Harness, maintained by Stefano's AI Lab. It is not affiliated with DeepSeek Inc.

@snnh Silver ? Unverified

Role presets and a model pool for dsh: the model delegates to a role, the operator's rules pick the model, and a failure walks the role's chain.

@kongbai9420 Gold ? Unverified

DeepSeek Harness 主从协作团队插件：主控架构统筹，子模型并行执行，闭环严苛验收

@noirabaka Gold ? Unverified

面向本地部署的 DSH 上下文压缩控制面：摘要改走指定模型、阈值真正生效、口径对齐卡片、暂停即完整还原

@shuangwang-cn Silver ? Unverified

Share your WorkBuddy credits with DeepSeek Harness, OpenCode, Cursor and other AI tools. Windows desktop app.

@residenthiago2011 Silver ? Unverified

Laya Tool Router Using the SystemOne Jev Typed Decisions for Local Laya Model and Equivalents.

@xuzihao66 Gold ? Unverified

Cut the input-token cost of long DeepSeek Harness sessions: source-side tool-output compression on tools/post-execute plus per-agent tool-catalogue visibility, with a measurement script.

@kee0012 Gold ? Unverified

科研技能中心 for DeepSeek Harness：技能目录、分类、按会话启用门控，以及用户可调用的 /command 技能栏。

@kee0012 Gold ? Unverified

DSH 专家团插件：把一支多角色专家团队装进当前会话 —— 团队画廊、成员名册、一键召唤，每位专家在独立子会话中并行工作，主会话汇总产出。

@baobao1996121 Silver ? Unverified

DSH 插件：ChatGPT 订阅登录、主 Agent 模型选择与直连/Clash 网络设置

@wuanthony397-hash Gold ? Unverified

Run the local Codex CLI as a peer agent from DeepSeek Harness: split work between the two agents, hand tasks over, review each other's changes, and keep every run on disk.

@fyue7 Silver ? Unverified

让 DSH agent 的输出自带情绪：会话投影情绪状态机 + 三段提示词注入（声线 / 状态 / 本轮实况）+ 会话头部情绪条，附文本蒸馏流水线（两轨蒸馏 / 保真门 / 输出护栏）。非官方插件。

@frankshi2024 Gold ? Unverified

Layered compaction-threshold manager for DeepSeek Harness: global / context-window tier / model policies with a live sidebar editor.

@majidasgharitabrizi Silver ? Unverified

Evidence-first agent memory and durable mission runtime — UNKNOWN states, immutable belief history, governed learning, and missions that survive kill -9. Zero-dependency TypeScript kernel.

@thefold060629 Gold ? Unverified

DSH 任务辅助启动器：在对话框发送键旁放置一个常驻按钮，点击后自动扫描本会话已安装的、对当前任务有帮助的技能与 MCP 工具，弹候选清单（默认勾选匹配项），一键确认后把选中技能注入当前对话交给 Agent 自动执行。

@mutoharohfiqhiabcd-source Gold ? Unverified

DSH 领导面板：与「领导」对话、切换它的模型、实时看 token/金额/预计剩余时间，并由它把任务派给工人 AI（Agent Teams）

@studyzy Gold ? Unverified

dsh有钱人才用得起的插件

@samirliu Gold ? Unverified

Deterministic goal-completion gate for DSH: contract-first AC grammar, re-runs every verify command itself (zero model trust), binds verdicts to a tree digest, denies false completion on update_goal/update_task via tools/pre-execute. Ports goal-loop's judging face onto native Agent Teams + task board.

@ruiyukirin Gold ? Unverified

DSH 输入框一键提示词增强按钮 | One-click prompt enhancement for the DeepSeek Harness composer: rewrite the unsent draft into a clearer, more specific prompt, restore the original with one click. Ported from Tencent WorkBuddy's EnhancePrompt design; independent model call that never touches the session log.

@wei894348 Gold ? Unverified

Self-hosted gateway plugin for DeepSeek Harness: registers a local WorkBuddy / CodeBuddy account pool as OpenAI-compatible model providers, with a web settings panel and a task-automation engine.

@ejdrone Gold ? Unverified

让 DSH 的一个会话由外部 ACP agent CLI 作为根 agent 驱动

@66wjz Gold ? Unverified

DSH 提供商与模型高级管理台 · Provider & model management console for DeepSeek Harness (dsh)

@magica-chen Silver ? Unverified

DeepSeek Harness agent preset: Codex and Claude Code as delegation subagents, each at a read-only and full-access tier. Use either — or both in parallel — in one task.

@fish-under-sea Silver ? Unverified

DSH 插件（上游 Ebony-Vinyl/dsh-our-free-model 的 Fork）：把 13 个免费渠道与免费模型并入官方模型选择器，自带渠道网关供其它本地 harness 反代；本 Fork 新增「隐藏渠道模型分组」总开关与按面积计价。

@lutisure Silver ? Unverified

昔涟角色 agent + 客户端主题 + 本地安全扫描，三个可独立安装的 DSH 插件。

@baroncyrus Gold ? Unverified

Use a PackyCode (PackyAPI) API key in DeepSeek Harness: built-in multi-protocol model route, ordered API-key rotation, and balance display

@j62268781-alt Silver ? Unverified

把 Hindsight 编码插件的 bank 身份锚定到远程仓库名，而不是本地目录名。Anchors Hindsight coding-agent memory banks to the remote repository name instead of the local directory name.

@dragonhorizon-xtihc Gold ? Unverified

把 **OpenCode Zen** 的免费模型接进 **DeepSeek Harness**：实时发现官方免费档模型，在设置里逐个启停，一键写入 provider。**免 key**

@vicqiufeeng Gold ? Unverified

Use the MiniMax Code (Mavis) subscription already signed in on this machine as a DeepSeek Harness model provider — with self-renewing credentials, a points/check-in panel and a model roster editor.

@mili1360 Silver ? Unverified

DSH plugin: folds the oldest part of a session history into one message backed by a lossless local archive. No model-written summary, everything recoverable.

@azazo1 Gold ? Unverified

把用户维护的文字块注入 DSH 系统提示词: 默认给出工具调用之间的进度汇报规范, 用来提高 agent message 输出频率

@defaver04 Silver ? Unverified

Free OpenCode Zen models for DeepSeek Harness (DSH) — native dsh-llm adapter, no API key. Fork with muse-spark unblocked, xhigh/max thinking levels, and a session-log safety guard.

@imzl-zl Silver ? Unverified

Transparent cross-provider failover routing for DeepSeek Harness agents.

@xinzeqing Gold ? Unverified

User-authored tiered long-term memory for DeepSeek Harness: injected as a system-prompt section, with a Chinese panel to view and edit it. The AI never writes to memory on its own.

@hyqhyq3 Gold ? Unverified

Bridge Google AI subscription (Antigravity/Gemini OAuth) into DeepSeek Harness as an OpenAI-compatible endpoint

@jonah791 Silver ? Unverified

DSH 多实例通讯底座：文件总线 + 在线名册 + 点对点/广播 + 收件箱 + 会话注入（多智能体工作台的 L0 层）

@linbin-mk Gold ? Unverified

deepseek-harness 工作区自定义提示词

@callqh Silver ? Unverified

Sign in to OpenAI Codex from DeepSeek Harness with a ChatGPT Plus/Pro subscription.

@zeranhub Silver ? Unverified

Interactive behavioral shaping for a DeepSeek Harness (DSH) session: rewrite the model-visible conversation history in place - edit a reply or its reasoning chain, delete a message, undo any rewrite, with an audit trail. Continuation of @wasd258/dsh-context-surgery (MIT, (c) 2026 WASD258-jpg).

@yunyv Silver ? Unverified

DSH 插件：生成式界面 / Generative UI for DeepSeek Harness。模型写声明式 DSL，无 DOM Worker 沙箱里执行，宿主用原生 DOM 画成真界面；另有一条自包含文档的逃生舱。原地改写、版本历史、跨重启可 patch。对标 ChatGPT Intelligent UI 与 Claude Artifacts。

@honwee Gold ? Unverified

Give your DeepSeek Harness agent a talking face: drive a Mynah realtime digital human (speak, interrupt, sessions, channels, knowledge base).

@lance808 Gold ? Unverified

DeepSeek Harness plugin: make the agent reason (think) in Chinese by contributing a system-prompt section

@snnh Silver ? Unverified

Community plugin for DeepSeek Harness: a lighter LLM service with self-contained provider profiles, a declared model catalog, and the chat-completions, Anthropic Messages, and OpenAI Responses wires. Not an official DeepSeek package.

@git4oc Silver ? Unverified

ZY-Agents 是单租户自托管的企业级多智能体平台，采用管理面与执行面分离的架构：主服务负责认证、编排与数据，内核宿主负责 Agent 运行时与本地工具执行。

@xfgil Gold ? Unverified

DSH 会话时钟：按会话授时 + 跨轮/跨会话空闲间隔，作为内置 dsh-time-context 的补充插件。Per-session clock injection for DeepSeek Harness with durable idle-gap tracking. A companion to the built-in dsh-time-context.

@suanlv Silver ? Unverified

DSH plugin: pool several local inference endpoints and spread delegated subagents across them

@h2o-yaoze Bronze ? Verified

Self-hosted mobile workbench built on DeepSeek Harness: projects, artifacts, code translation and usage.

@caoran1314 Gold ? Verified

三省六部制多智能体审核流程，移植为 DeepSeek Harness 插件：12 状态状态机、门下省封驳、高风险确认、越权矩阵、委派防死锁、完整审计、军机处看板。

@ztllll Gold ? Verified

中文优先：让滴答清单成为人类与 Agent 共享的 Todo 真源（Pi 扩展 + dsh 插件），双向同步、提醒、强制人类验收，dsh 会话中断自动续跑。

@sssogin Silver ? Verified

Multi-Agent Long-Task Scheduling and Growth System for AI coding agents, with file-based state, verification, handoff, and reusable learning.

@zariaecho Silver ? Verified

GitHub deep-workflow tools for DeepSeek Harness (DSH) agents: issue triage with 4-layer context, template-aware draft PRs, graded PR review, CI diagnosis and GraphQL merge — behind a fail-closed mutation guard (read-only / approval / token scopes).

@xp266 Silver ? Verified

ink-based terminal UI plugin for DeepSeek Harness

@huangfuren Gold ? Verified

DSH web plugin: search, read and safely write an Outline knowledge base from conversations, with whitelist-guarded approval for every document write.

@wayaero Silver ? Verified

DeepSeek Harness 的 ESP32-S3 Buddy 插件：通过 BLE 显示会话、Token 与上下文状态，支持触摸审批、电脑校时和角色包传输。

@jiale-li-orion Silver ? Verified

One agent across many devices — a multi-device capability runtime and personal workbench for persistent agents, built on DeepSeek Harness.

@manoloremiddi Gold ? Verified

Reusable prompts in DeepSeek Harness Settings, with standalone SQLite storage or an existing Augmentor service.

@nbagent-dev Gold ? Verified

Meta-Problem Modeling (MPM) cognitive flywheel as a resident engine for LLM agent runtimes. 六阶段问题生命周期 · ε-δ 双判据收敛 · 代谢账本 · 导师规则层

@01men Gold ? Verified

榕器定制轨（01门）· AI 代理与人类协作前台——行业 AI 工作台（IAW）与一线面板/看板：dsh 一键插件化（班组长工作台/执行卡/Skill 点名直调/钉钉桥），上游 01men/ybkk-AIOS

@leolee9086 Bronze ? Verified

DeepSeek Harness 的疲劳度、唤醒值与自主上下文压缩 Cordis 插件

@john-walks-slow Gold ? Verified

Humanized fork of @xmanrui/dsh-im (base 4.21.2): humanized message delivery with send-delay + typing two-phase, message_break + streaming coexistence, status reactions, reply quotes, per-bot overrides and a no_reply reclaim tool for DeepSeek Harness IM channels.

@ravenwangchina Silver ? Verified

Meta-Forge Workshop: turn recurring work into governed AI workflow assets. MCP server + Claude Code plugin + dsh plugin — staged writes, evals-gated forging, inference engine. 元铸工坊 SOSOFAST：把最佳实践铸成可治理的 AI 工作流资产

@funcpn Silver ? Verified

Keeps a project's design doc in step with its decisions instead of letting it drift: an append-only decision log, a design snapshot derived from it, and a hook that refreshes it automatically. ADR-style docs that stay fresh. Ships as a plugin for Claude Code, Codex and DeepSeek Harness.

@dsh-eac ? Verified

EAC 与 AIO 自定义皮肤的 AI Prompt 包、来源清单与 Schema

@mario841859784 Silver ? Verified

DSH 专家编排模式 agent preset — PM-first planning, Agency expert delegation, gated delivery, experience pooling

@lisonevf Bronze ? Verified

在 AI 会话窗口里直接生图 / 改图 —— dsh 插件，本地跑 Qwen-Image-2.1（原生 2K / 原生透明 PNG / 最多 10 张参考图）。安装器自动适配 CUDA 与已有 ComfyUI/conda 环境

@fish-under-sea Silver ? Verified

DSH 插件（上游 NanmiCoder/dsh-agent-teams 的补充版）：AgentTeams 多智能体团队协作 —— 自然语言组队、任务依赖 DAG、信箱通信、右侧栏树状监测；随包自带 17 厂商 × 10 岗位头像与 17 个厂商徽标（286 个文件），支持自定义美术目录与中文岗位名。

@panando Silver ? Verified

One button prompt optimizer with configurable strategy.

@vitas Gold ? Verified

Cut LLM costs: route routine tasks to cheap subagent models (DeepSeek Harness plugin)

@hercules-debug Gold ? Verified

把「我改完了」变成可验证的事实:Δ(预期改动)+ P(检测程序)全过才产生 commit。零依赖,只用 node 和 git。 | Turn 'done' into a verifiable fact — a commit exists only if Δ + P both pass.

@windwhiterain Silver ? Verified

A persistent office of colleague agents for DeepSeek Harness: ordinary sessions with a roster, channels, direct messages, and delivery that wakes a colleague when something is addressed to it.

@wayaero Bronze ? Verified

ESP32-S3 桌面 AI 伙伴：触摸审批、工作状态、角色动画与蓝牙校时，支持 DeepSeek Harness 和 Claude Code。

@amadeus2027 Silver ? Verified

Visual sketchboard for DeepSeek Harness: draw with Excalidraw, let AI agents inspect and annotate sketches, and review proposed edits.

@2026heshao Silver ? Verified

All-in-one board for DeepSeek Harness: MCP manager + Skills + WorkBuddy experts (attach/switch/detach any time) + tool-usage tailor + three direct markets (WorkBuddy connectors, SkillHub 190k+ skills) with one-click install · DSH 一体化看板：MCP/技能/专家随时挂换/工具账本/三大市场直连安装

@hercules-debug Gold ? Verified

Self-organizing multi-agent clusters for DeepSeek Harness — real DSH subagents coordinating through a shared append-only log, no orchestrator

@azetus Silver ? Verified

Floating desktop chat panel for DeepSeek Harness — global-hotkey access to sessions, models, and tool approvals while you work in other apps. MacOS/Windows.

@jmx-4869 Gold ? Verified

给 DeepSeek Harness 加一层心情。33 种心情（基础 / 复合 / 当代）各自持有数值，每 3~6 小时自己变 —— 它会累、会烦、会好奇。附带每日心情日记、时事、可选的主动开口。MIT 开源。

@lengfengng-tech Gold ? Verified

Per-workspace persistent memory plugin for DeepSeek Harness (DSH): sidebar panel, automatic system-prompt injection, and an agent tool to save memories

@xiamubumu Gold ? Verified

DSH 1.5 启动器：一个面板统管能力组合、插件清单、MCP 桥与浏览器标签页扩展。仅支持 DSH 1.5，后续会跟进更新。 / Launcher for DSH 1.5: one panel for the composer, the plugin inventory, the MCP bridge and the tab-bridge extension. DSH 1.5 only; newer releases will follow.

@1015677515 Gold ? Verified

deepseek harness 全局提示词/对话提示词，错误记录功能插件