
Patent Disclosure Skill Adds Smarter Workflows
This MIT-licensed Agent Skill turns project documents, source code, and patent PDFs into Chinese technical disclosures, plain-language patent notes, Obsidian knowledge graphs, prior-art searches, and draft office-action responses. Recent commits add classification-based searches, response-strategy routing, and automated structural line-art assembly.
The project’s strongest idea is treating patent preparation as a repeatable agent workflow instead of a one-off prompt, giving engineering teams a practical bridge from implementation to intellectual-property documentation.
- –Scans DOCX, PPTX, Markdown, code, and design materials, then produces Markdown and Word deliverables with diagrams.
- –CNIPA-focused prior-art search and classification filtering add useful structure, but automated novelty checks should not replace professional patent searching.
- –Obsidian integration, revision logs, and optional RAG turn isolated patent work into reusable organizational knowledge.
- –Support for invention, utility-model, and design patents makes the workflow unusually broad for an Agent Skill.
- –The project is China-specific, Chinese-language, and dependency-heavy; its recent activity reflects iterative expansion rather than a brand-new release. [Commits](https://github.com/handsomestWei/patent-disclosure-skill/commits/main) [Limitations](https://labgrimoire.com/spellbook/patent-disclosure-skill/)
DISCOVERED
1h ago
2026-08-30
PUBLISHED
1h ago
2026-08-30
RELEVANCE
