1. Conduid
  2. Social
  3. Deepseek Harnessed
MCP server · Social

Deepseek Harnessed

Turn a DeepSeek Harness (DSH) instance into a subagent any harness (Cursor / Claude Code / Codex / Gemini CLI ...) can call over MCP - plus a live monitor panel inside DSH Desktop.

34Low

Scored a month ago · breakdown

About Deepseek Harnessed

Deepseek Harnessed is an MCP server published by xDJTomato in the Social category: turn a DeepSeek Harness (DSH) instance into a subagent any harness (Cursor / Claude Code / Codex / Gemini CLI ...) can call over MCP - plus a live monitor panel inside DSH Desktop.

Recent commit activity is the strongest signal that reported issues will get attention.

Install

From source
git clone https://github.com/xDJTomato/deepseek-harnessed

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

✦

Ask AI about Deepseek Harnessed

Powered by Claude · Grounded in docs

I know everything about Deepseek Harnessed. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

6 checks have not been run yet.

Releases

v0.1.11v0.1.11 —— 面板不再拖着宿主 GPU 一直烧;顺带发三条积压的委派规则 · 29 Sep 20260.1.11 —— 面板不再拖着宿主 GPU 一直烧,顺带发三条积压的委派规则 1) 修复:子代理面板与设置窗同时打开时界面卡死 现象是你报的:两个窗口一起开,界面非常卡。根因是两件事相乘 —— 卡片上那层缓慢扫过的光泽是常驻无限动画。这张卡片带 mask-image 的伪元素 + overflow: hidden,走不了纯合成路径,宿主只能按帧率重绘整张卡片; 设置窗那层是全视口 backdrop-filter —— 背后内容一变就必须重新模糊。…
v0.1.10v0.1.10 —— 预览窗口只读显示子代理的对话内容 · 21 Sep 20260.1.10 —— 预览窗口里只读显示子代理的对话内容 1) 新增:只读会话转录 运行中的外部任务不能点开(点开 = 宿主接管会话写权、写坏子进程正在写的日志),原先那个只读 详情页只能看到 stderr.log 的最后几行,看不到它到底在说什么。现在观察器在算 Token 用量的 同一趟解码里把会话日志折成转录,随投影下发给卡片,详情页按四类角色只读渲染: 用户 / 助手:取消息体里的文本块;推理块(reasoning)不显示 —— 长且刷屏;…
v0.1.90.1.9 — 文档重写为朴实技术语言 + 公开仓库脱敏 · 21 Sep 20260.1.9 —— 文档改写为朴实技术语言 + 公开仓库脱敏 这个版本没有新功能,改的是文档本身,以及文档里本不该公开的信息。 1) 四份标准文档重写 README.md、docs/clients.md、docs/configuration.md、docs/install.md 用 gemini-3.7-flash 重写为 朴实技术语言:把"磁贴 / 硬承诺 / 响亮失败 / 口径 / 幽灵 / 垫片 / 探针 / 闸门"这类生造说法 换成常见词(卡片 / 明确的执行上限 /…
v0.1.80.1.8 — 模型选择:首次接入引导 + 用户可编辑策略文件 · 21 Sep 20260.1.8 —— 模型选择改成"首次接入 + 可编辑策略文件" 这个版本解决一个真实问题:以前桥接层把"可用模型"写死在工具描述里,换机器、换 DSH 实例、换了网关,描述立刻变成错的,而调用方会照着错描述去传参数。 改动: 1) 首次接入引导 策略文件里还没指定默认模型时,dsh_task 与 dsh_health 的返回值末尾会带一段紧凑引导: 先枚举这个 DSH 实例已接入的模型 -> 请用户指定一个默认模型 -> 用 dsh_setup 落盘 ->…
v0.1.7v0.1.7 — dsh_task 可按次指定模型/服务商/推理强度;写错即响亮失败 · 20 Sep 2026## [0.1.7] — 2026-09-20 新增 `dsh_task` 暴露 `reasoning_effort`(单次调用改推理强度):`lib/tasks.mjs` 的 `startTask()` 早就支持它(写进 `task.json` 的 `reasoningEffort` 并拼出 `--reasoning-effort <值>`),但 `lib/mcp.mjs` 完全没有暴露这个参数。现在 schema 里有它(`type:…
Questions

About Deepseek Harnessed

How do I install Deepseek Harnessed?

No package has been published for this server, so install it from source: clone the repository and follow its README.

Is Deepseek Harnessed safe to use with an AI agent?

Its trust score is 34 out of 100 (low). Conduid hasn't run static security checks on this repository yet, so review the source yourself before granting it credentials. It has no ConduID identity yet, so agent calls to it are not receipted.

Is Deepseek Harnessed still maintained?

Yes — the latest release is v0.1.11 (29 Sep 2026), and the last commit was 19 days ago. The repository has 0 stars.