Third-party DSH plugin for patent work: Chinese-patent triage with publication-number evidence, eight-chapter disclosure drafting, drawio figures, Docker experiments, a seven-dimension review and docx/PDF export.
DSH integration
Native runtime
Author-claimed
Safety audit
Unaudited
Last verified
2026-11-21
License
Not specified by the author
01What can it help you accomplish?
Decide whether a technical idea is worth a patent application before anyone writes a draft
A three-tier verdict — write it, narrow it first, or do not write it — with Chinese patent publication numbers as evidence, then a five-dimension interview that checks the drawback–problem–effect chain lines up
Inventors, patent engineers and R&D leads who want a go/no-go call before spending weeks on a disclosure
Turn an approved direction into a disclosure a patent attorney can work from
An eight-chapter disclosure — the default deliverable — with a body-text check and an effect comparison, plus a verification chapter whenever experiments are involved; claims, specification and abstract only on explicit request
Engineers who have to hand over a structured disclosure instead of a pile of notes
Finish the whole package — figures, experiments, review, export — without leaving DSH
drawio or HTML architecture and structure figures with geometry and figure-text consistency checks, Docker-simulated experiments with run logs, a seven-dimension review report under `review/`, and docx/PDF export of the disclosure and filing documents
Teams that want drawings, simulation and export in one pass instead of a separate drawing, simulation and typesetting toolchain
02How to install into DeepSeek Harness
Prerequisites
- A compatible dsh host with an available model — the README lists both as runtime requirements.
- Network access to `patents.google.com` for the Chinese patent search.
- Docker for simulated experiments; a local drawio CLI or Docker for drawio figures; Edge or Chrome for HTML figure rendering; for PDF, Windows Word COM or a configured weasyprint environment.
Installation steps
- 01
DSH Desktop: open **Settings → plugin marketplace**, search **Aurify** or **点金**, and install it into the active profile — no repository clone or release download needed.
- 02
In that profile send the setup prompt: 帮我初始化点金插件的运行环境,检查并配置所需依赖和服务。
- 03
Work through the `patent_setup_check` report, and after any MCP configuration change fully quit and restart the desktop app.
- 04
Send the recheck prompt 帮我复检点金插件的环境,确认所需服务已就绪。 then describe your technical idea, or hand over an existing Word document to keep refining it.
Verify the integration
- `patent_setup_check` returns a report showing the required dependencies and services as configured.
- The recheck prompt comes back clean — the README says that once the recheck passes you can describe a technical idea and start the assessment.
03DSH integration and capability boundaries
Native DSH plugin: install it from the desktop Settings → plugin marketplace (or the community DSH Market) into the active profile, and it lands as 15 skills, 5 native tools, 15 Python MCP tools, a standalone review command, Web cards and a project panel.
Patent triage and alignment interview
A technical idea described in chat, or an existing Word document→One of three verdicts (write / narrow first / do not write) with publication-number evidence, then five dimensions of information and a drawback–problem–effect alignment check
queries `patents.google.com` over the network for the Chinese patent searchEight-chapter disclosure drafting
The confirmed direction plus what the interview collected→An eight-chapter disclosure with a body-text check and an effect comparison, and an extra verification chapter when experiments are involved
Figures and simulated experiments
The structure and flow descriptions in the disclosure→drawio or HTML architecture and structure figures (cutaway and exploded views included) with geometry checks and figure-text consistency acceptance, plus Docker-simulated experiments and their run logs
renders figures through a local drawio CLI or Docker, or through Edge/Chrome on the HTML routeruns the simulated experiments inside DockerSeven-dimension review
The drafted disclosure→A score along seven fixed dimensions written to a report under `review/`, reachable from both the `/patent-review` command and the `patent_review` tool
writes the review report into the project's `review/` directoryWord import and docx/PDF export
An existing Word document, or a finished disclosure and filing documents→Imported paragraphs turned into project chapters, and docx/PDF files for the disclosure and the application documents
writes docx/PDF output filesPDF export prefers Windows Word COM and otherwise needs a configured weasyprint environment
04Who is it for? When not to use it?
Good for
- Inventors, patent engineers and R&D leads who want a go/no-go call before spending weeks on a disclosure
- Engineers who have to hand over a structured disclosure instead of a pile of notes
- Teams that want drawings, simulation and export in one pass instead of a separate drawing, simulation and typesetting toolchain
Not for
- The patent search needs reachable `patents.google.com`, simulated experiments need Docker, drawio figures need a local CLI or Docker, HTML figures need Edge or Chrome, and PDF export prefers Windows Word COM or a configured weasyprint environment — none of these ship with the plugin.
- The review process is fixed and model scoring still fluctuates, so passing the checks is no grantability guarantee; Word import only extracts paragraphs, so tables and embedded images have to be checked by hand.
05Compatibility, maintenance and safety notes
- The patent search needs reachable `patents.google.com`, simulated experiments need Docker, drawio figures need a local CLI or Docker, HTML figures need Edge or Chrome, and PDF export prefers Windows Word COM or a configured weasyprint environment — none of these ship with the plugin.
- The review process is fixed and model scoring still fluctuates, so passing the checks is no grantability guarantee; Word import only extracts paragraphs, so tables and embedded images have to be checked by hand.
- The disclosure is the only default deliverable — claims, specification and abstract appear only when you ask for them — and the bundle carries no persona; a dedicated persona is an optional profile-level install.
- GitHub Releases, npm/PyPI and the plugin marketplace publish independently, so their versions can drift; `mcp_wheel: true` pulls an unpinned PyPI package, so use source mode to run exactly this version of the Python service.
No license stated in the README · actively maintained (latest release v0.3.0, 2026-11-21; npm packages and Python service both at 0.2.0)
06Frequently asked questions
How do I install Aurify in DeepSeek Harness?
In DSH Desktop open Settings → plugin marketplace, search Aurify or 点金 and install it into the active profile — no clone and no release download are needed. The plugin is also listed in the community DSH Market, so you can install it from that sidebar instead.
What do I do right after installing it?
Send the setup prompt in that profile (帮我初始化点金插件的运行环境,检查并配置所需依赖和服务。), work through the `patent_setup_check` report, then fully quit and restart the desktop app after any MCP configuration change and send the recheck prompt.
What does it look like inside DSH once installed?
Fifteen skills, five native tools and fifteen Python MCP tools, plus Web cards and a project panel. You drive it from chat, trigger the review with `/patent-review`, and read the resulting report from the project's `review/` directory.
Which DSH version or platform does it need?
A compatible dsh host with an available model. Beyond that the README only lists external needs: `patents.google.com` for the patent search, Docker for experiments, Edge or Chrome for HTML figures, and Windows Word COM or weasyprint for PDF.
How does it sit next to other DSH plugins?
It is a standalone third-party plugin with no affiliation to DeepSeek Harness. It installs from the marketplace like any other plugin, ships no persona, and touches nothing in the profile beyond its own cards and panel.
07Related DSH workflows
dsh-market
by dsh-market
插件市场 (plugin marketplace): 打开 Settings → Plugin Market 浏览/搜索/一键安装社区插件;安装源限制为 curated awesome-dsh-plugin 注册表(其余拒绝),默认禁用构建脚本(pnpm>=10 需显式开启)。
dsh-guizang-design
by op7418
Expose five content/design skills, a gallery, and multi-skill workflows inside DSH.
oh-story-dsh
by zenstory-ai
Fiction/short-drama writing workbench: 23 skills, roles, markdown/JSONL preview, live file-follow.
dsh-skills-manager
by michengai
DSH 技能集合管理器,统一管理多个 skill。
08Data and sources
DeepSeek Harness(dsh)的第三方专利撰写插件。
插件由 15 个技能、5 个原生工具、15 个 Python MCP 工具和独立审查命令/工具组成,包含 Web 卡片和项目面板。
在 DeepSeek Harness 桌面版 **设置 → 插件市场**中搜索 **Aurify** 或 **点金**,安装到当前活动档案。
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-11-21. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.
