What does this skill do, and when should you use it?
This skill converts papers, theses, technical reports, source code, figures, inventor notes, or research manuscripts into Chinese invention patent drafts and attorney-facing technical disclosure materials. It first loads the workflow definition (manifest.yaml) and permanent files, detecting source format, task mode, and invention type. It then creates stable source IDs for paper text, equations, figures, and code, ensuring every claim feature maps to explicit evidence. The skill progresses through stage gates: building a source map, terminology ledger, and evidence inventory before drafting claims and specification. Final deliverables include Chinese DOCX files with native editable Office Math formulas and Mermaid diagrams.
Runs Python workflow scripts (validate_patent_draft.py, build_patent_package.py) to generate Chinese DOCX; extracts tables and formulas (embedded as native Office Math); renders flowcharts and system diagrams via Mermaid; performs internal consistency checks and produces validation reports; outputs claims, specification, abstract, figures, and technical disclosure in Markdown and DOCX.
- A researcher converts a published paper into a Chinese patent draft with systematic claims and evidence mapping.
- An inventor combines lab notes and code to write a well-grounded technical disclosure for an attorney.
- A patent engineer compares an existing paper against a patent for novelty and consistency audit.
- An R&D team mines patent points from project materials and generates preliminary disclosure documents.
- A researcher needs to document an algorithmic invention with core formulas as editable Office Math in DOCX.
How do you install this skill?
- The skill depends on overseas services (e.g., CNIPA search requiring Playwright and external network); access from mainland China may be restricted and core features may not be fully usable.
- Publisher is unverified; maintenance responsibility and update path are unclear. Verify recent commits and issue responsiveness before use.
- The output is a drafting aid for inventor and professional review, not a legal opinion; always have a patent attorney validate.
- The skill may execute scripts (e.g., validation, package building). Run in a controlled environment and review script behavior.
- Shell / CLI
- Network access
- Local filesystem
Python 3python-docxmermaid-cli
Install the full collection via npx or by cloning the repo (e.g., npx skills add ...). This skill requires Python 3 with additional packages (requirements.txt) and Mermaid rendering; for CNIPA search, also install Playwright.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/Yuan1z0825/nature-skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/nature-paper-to-patent" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Turn this paper into a Chinese patent draft
In a compatible agent, describe the paper or materials to convert (e.g., "Turn this paper into a Chinese patent draft"), and the skill will trigger and run the full workflow automatically.
What are this skill's strengths and limitations?
- Enforces evidence grounding, making claims traceable to specific sources.
- Supports multiple input formats (PDF, scanned PDF, text, mixed project).
- Produces complete DOCX patent packages with editable Office formulas.
- Includes validation scripts to catch errors and warnings.
- Skill is in Beta, may have edge cases on real-world inputs.
- Requires Python environment and dependencies, increasing setup bar.
- Generated drafts are not patentability opinions and require professional review.
- Full automation from paper to patent may be slow and require iterations.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Paper to Chinese Patent Draft this page | 56 · Use with care | ★ 47k | 3d ago | Apache-2.0 |
| OfficeCLI Academic Paper Skill | 51 · Use with care | ★ 32k | 5d ago | Apache-2.0 |
| Shuorenhua: Chinese AI-Tone Cleanup Skill | 65 · Recommended | ★ 2k | 11d ago | MIT |
| Offer & Claims Registry | 60 · Recommended | ★ 2.9k | 3d ago | Apache-2.0 |
| Resume Tailoring Skill | 51 · Use with care | ★ 767 | 7mo ago | MIT |
How did FollowSkills review this skill?
Evidence shows the skill mandates stable source IDs, marks unsupported features as 'unsupported' and excludes them from formal claims, does not infer inventorship/ownership/legal sufficiency, and asks users to confirm missing facts. These reflect data-flow transparency and user confirmation. However, no explicit least-privilege permissions, rollback, or external-effect control is documented, and the publisher is unverified. Deductions applied.
The skill defines a validation workflow (validate_patent_draft.py and build_patent_package.py) and has a tests directory with pytest in CI, but static review cannot execute these tests, and no explicit error-handling or failure-feedback design is visible. Cap at 10; given tests exist but no independent reproduction evidence, score 7.
The skill clearly defines use cases, typical requests, outputs, and boundaries, and supports Chinese outputs for Chinese users. However, core functionality may depend on overseas services (e.g., CNIPA search requiring Playwright and network), potentially unreachable from mainland China, and environment compatibility is not fully detailed. Score 10.
Documentation is well-structured with bilingual READMEs, follows information architecture largely, provides dependencies, examples, and boundaries. But hidden assumptions (Python dependencies, Playwright), no version history or changelog, and unclear maintenance responsibility (unverified publisher) lead to deduction.
The skill specifies detailed outputs (claims, specification, abstract, etc.) and an end-to-end process, but static review cannot verify output quality. Core task is clear and validation is specified, yet no execution evidence; cap at 7, score 6.
A CI workflow runs pytest for this skill, providing some third-party execution evidence, but test contents are not provided and independent reproduction is not possible. Cap at 5, score 4.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →