What is this?
An AI-agent skill for Chinese patents: it mines patentable points from your project materials and writes disclosure documents (invention / utility-model / design), rewrites disclosures into application files, runs bibliographic search, explains published patents in plain language, and helps draft examination responses.
An AI-agent skill for Chinese patents: it mines patentable points from your project materials and writes disclosure documents (invention / utility-model / design), rewrites disclosures into application files, runs bibliographic search, explains published patents in plain language, and helps draft examination responses.Is this a standalone app, or a skill my AI assistant loads?It is not an app with its own UI — it is a skill package (prompts, templates, sub-skills and Python scripts) that an AI agent loads and drives from chat, using triggers like '交底书' for disclosure writing or '读专利' for reading a patent.
It is not an app with its own UI — it is a skill package (prompts, templates, sub-skills and Python scripts) that an AI agent loads and drives from chat, using triggers like '交底书' for disclosure writing or '读专利' for reading a patent.How do I get started with it?Install the skill into your AI agent via the repo's detailed installation guide, then drop project materials into the chat: it mines the protectable points, checks prior art and outputs a submittable disclosure, with separate template sets for invention, utility-model and design cases.
Install the skill into your AI agent via the repo's detailed installation guide, then drop project materials into the chat: it mines the protectable points, checks prior art and outputs a submittable disclosure, with separate template sets for invention, utility-model and design cases.Do I need Obsidian for everything to work?Obsidian matters most for the reading side: each interpreted patent becomes linked notes, graphs and canvases in your vault, so reading grows into a personal patent knowledge base. The install guide covers Obsidian itself; only some community plugins are optional.
Obsidian matters most for the reading side: each interpreted patent becomes linked notes, graphs and canvases in your vault, so reading grows into a personal patent knowledge base. The install guide covers Obsidian itself; only some community plugins are optional.What happens when my materials are missing facts — does it just make things up?It is built not to fabricate: in the combined disclosure-and-application workflow it role-plays disclosure engineer vs patent attorney for at most three checklist rounds, asks you for missing facts, and moves unclear points into a question list instead of blocking the whole file set.
It is built not to fabricate: in the combined disclosure-and-application workflow it role-plays disclosure engineer vs patent attorney for at most three checklist rounds, asks you for missing facts, and moves unclear points into a question list instead of blocking the whole file set.Is it tied to Chinese patents, or does it work for other jurisdictions?Chinese practice only: invention, utility-model and design filings under CNIPA rules, office-action responses judged by Chinese examination standards, and policy briefs from the CNIPA website. It is written in Chinese for engineers who build tech but have never filed a patent.
Chinese practice only: invention, utility-model and design filings under CNIPA rules, office-action responses judged by Chinese examination standards, and policy briefs from the CNIPA website. It is written in Chinese for engineers who build tech but have never filed a patent.Which systems does it run on?Wherever Obsidian and your AI agent run — Windows, macOS and Linux; the repo is tagged for all three and the install guide details Windows specifics.
Wherever Obsidian and your AI agent run — Windows, macOS and Linux; the repo is tagged for all three and the install guide details Windows specifics.Where does my confidential disclosure material end up?In your own vault: interpreted patents accumulate in a private, personal Obsidian knowledge base rather than a shared service. The only outside source mentioned is the CNIPA site, which the policy sub-skill reads for recent announcements.
In your own vault: interpreted patents accumulate in a private, personal Obsidian knowledge base rather than a shared service. The only outside source mentioned is the CNIPA site, which the policy sub-skill reads for recent announcements.Is it free to use?Yes — MIT licensed, with only an optional coffee-donation request from the author.
Yes — MIT licensed, with only an optional coffee-donation request from the author.What does each sub-skill actually do?Seven: patent-disclosure drafts disclosure documents from raw materials; patent-application converts them into claims, description, abstract and black-and-white figures; patent-docket runs a role-play session producing both at once; patent-reader turns a publication number or PDF into plain-language notes and maps; patent-oa drafts office-action responses with RAG; patent-search does bibliographic lookup; patent-exam-policy briefs on policy shifts.
Seven: patent-disclosure drafts disclosure documents from raw materials; patent-application converts them into claims, description, abstract and black-and-white figures; patent-docket runs a role-play session producing both at once; patent-reader turns a publication number or PDF into plain-language notes and maps; patent-oa drafts office-action responses with RAG; patent-search does bibliographic lookup; patent-exam-policy briefs on policy shifts.Why did the author build this?After years of core R&D, the author was never listed as an inventor — the bottleneck was the writing itself: how to mine patent points, word the novelty section, and get diagrams plus an editable Word file submitted in one pass. The skill closes that gap for the people who actually build the technology.
After years of core R&D, the author was never listed as an inventor — the bottleneck was the writing itself: how to mine patent points, word the novelty section, and get diagrams plus an editable Word file submitted in one pass. The skill closes that gap for the people who actually build the technology.Can it handle images and drawings, or text only?Visuals are first-class: it reads structure and design drawings into disclosures, extracts design line art from product photos, generates utility-model outlines with component-number callouts from diagrams, and pulls multi-view projections from CAD 3D models — and patent-search can even reverse-derive a search query from a product image or a claim.
Visuals are first-class: it reads structure and design drawings into disclosures, extracts design line art from product photos, generates utility-model outlines with component-number callouts from diagrams, and pulls multi-view projections from CAD 3D models — and patent-search can even reverse-derive a search query from a product image or a claim.How does it keep up with patent-office changes?patent-exam-policy reads recent announcements on the CNIPA official site and turns them into plain-language briefs, flagging which disclosure techniques or application forms in the skill may have gone stale — so your documents stay aligned with the current examination line.
patent-exam-policy reads recent announcements on the CNIPA official site and turns them into plain-language briefs, flagging which disclosure techniques or application forms in the skill may have gone stale — so your documents stay aligned with the current examination line.What are similar projects?ljg-skills, dbskill, OpenKB, SenseNova-Skills, chubbyskills, ai-aug-edu, khazix-skills, axton-obsidian-visual-skills
ljg-skills, dbskill, OpenKB, SenseNova-Skills, chubbyskills, ai-aug-edu, khazix-skills, axton-obsidian-visual-skillsWhich topics is it in?legal-tech, obsidian, writing, chinese-language, knowledge-management, note-taking, document-processing, search, knowledge-base, skill, pdf, rag
legal-tech, obsidian, writing, chinese-language, knowledge-management, note-taking, document-processing, search, knowledge-base, skill, pdf, rag