DeepSeek Harness Plugin Hub

发布与管理完整 Harness Profiles,发现适合你的插件。

探索

插件目录环境预设文档中心动态

社区

发布插件联系我们报告问题

相关链接

Plugin Hub GitHubDeepSeek Harness 官方项目系统状态隐私说明
© 2026 DeepSeek Harness Plugin HubPowered byPaxTech

独立、非官方社区项目,与 DeepSeek 官方无隶属、授权或背书关系。

DeepSeek Harness Plugin Hub
ProfilesPlugins分类动态文档登录管理 Profiles
ProfilesPlugins分类动态文档登录
← Plugins
P

dsh-penpot

Penpot

适用于 DeepSeek Harness 的 Penpot 设计平台集成:提供 REST/RPC 工具(不使用 MCP),并附带设计评审和交付技能。支持 penpot.app SaaS 或任何自托管的 Penpot。

插件会安装到这里;不确定时保持 web。

npx -y @deepseek-ai/dsh plugin --profile web add github:AlbertZJ/dsh-penpot#3305bfe7e6483378eced94f2cf55f023ecfaa7fd
README兼容性版本

兼容性与来源证明

Penpot 以 dsh-penpot 发布,当前版本为 0.2.0。Plugin Hub 会校验它的 manifest,并保存精确安装来源,便于复现安装结果。

DSH 兼容范围
*
运行环境
any
发布来源
github
Registry 更新时间
2026/10/24

版本

0.2.0stable
2026/10/24

相关插件

正在加载相关插件…

最新版
0.2.0
DSH
*
HMR
重启进程
Tree shaking
未声明可安全裁剪
解包体积
未提供
文件数
未提供
Surface
any
许可证
MIT
发布源
github
GitHub
★ 0
周下载
0
最近提交
2026/10/24
查看源码 ↗
README Badge

点击下方 Badge 复制 Markdown,粘贴到 README 即可。

这是你的 Plugin?认领权益 · 优先安全扫描

验证 package.json 声明的 GitHub 仓库,即可管理这个公开页面。认领后,Hub 会优先安排当前版本的安全扫描,并在通过后公开展示结果。

认领这个 Plugin →
报告问题

相关插件

继续浏览 integrations-communication 分类下经过校验的插件。

Acp App@deepseek-ai/dsh-acp-appdsh ACP 配置文件包:基于 dsh-base 的仅限自动化的 JSON-RPC stdio 和进程生命周期管理Remote Web Ui@linxin666/dsh-remote-web-ui通过扫码配对访问 dsh Web GUI,共享一个官方界面:设置按钮旁的二维码可将手机和 PC 配对到同一个 Web GUI(手机采用竖屏触控适配层,PC 使用完整桌面界面),通过一次性令牌和 rRemote Bridge@linxin666/dsh-remote-bridgeDSH 远程桥接:主机端通过唯一且精确的 /ios/v1 WebSocket 路由提供协议 v1——会话列表、历史记录基线、提示、包含工具活动的实时回合增量、中断以及进程内代理审批——连接到通过一次性配对的设备Ds Botds-botDS Bot:DeepSeek Harness 的一组长期运行的 Bot,支持主 Bot、Bot 间消息和群聊

README

dsh-penpot

Penpot inside DeepSeek Harness as a native plugin — no MCP. The plugin registers model-facing tools that call Penpot's backend command API directly, plus two bundled skills (design review and design handoff) written against those tools.

Works with penpot.app SaaS or any self-hosted Penpot.

English | 中文

What it does

  • 13 penpot_* tools over Penpot's REST/RPC command API, read and write: connectivity/profile check, teams → projects → files walk, file search, file overview (pages + design-token sets), page board tree, open comment threads, binary .penpot archive export; plus create project, create file, import .penpot archive, and upload generated SVG/PNG/JPEG/WebP assets into a file. Authentication uses a personal access token via the Authorization: Token scheme — the same backend API Penpot's own web app uses.
  • 3 bundled skills: penpot-design-review (structured design review from live data), penpot-design-handoff (inventory + tokens + comments + portable archive for developers), and penpot-design-generate (create a project/file and deliver model-authored SVG boards as Penpot assets — the write path, no MCP, no canvas automation).

Requirements

  • DeepSeek Harness >= 0.2.1-rc.2 < 0.3.0
  • Node.js >= 22.19.0
  • A Penpot personal access token: Penpot → Settings → Access tokens → generate one.

Install

export PENPOT_TOKEN=<your-token>            # Windows PowerShell: $env:PENPOT_TOKEN="..."
dsh plugin --profile web add github:AlbertZJ/dsh-penpot

Optional configuration (defaults shown) — edit the bundle row's config on the Plugins page, or in cordis.patch.yml for linked installs:

config:
  baseUrl: https://design.penpot.dev   # or your self-hosted origin
  tokenRef: PENPOT_TOKEN               # env var holding the token
  timeoutMs: 30000

No build permission is needed on install: lib/ is committed and the package has no prepare script.

Verify

In a session, ask the model to run penpot_whoami — it should return your Penpot profile. Then try "review the design draft named X" and the penpot-design-review skill takes over.

Development

pnpm install
pnpm run build     # tsc + tsdown -> lib/
pnpm run verify    # structural checks (no dependencies)

lib/ is committed: git installs fetch sources and do not build them, so rebuild and commit lib/ together with any src/ change.

How it compares

  • Unlike qunanchang/dsh-penpot, this plugin does not embed the Penpot editor UI into the sidebar and does not use Penpot's MCP. It is a data/tool integration: the model reads and exports design data through dsh-native tools.
  • The two approaches compose well: use dsh-penpot for agent workflows, and the other one if you want the visual editor inside dsh.

License

MIT