SuperAPI AI Directory

patent-disclosure

Writing & research

View repository

Updated 10-15 22:12

About this skill

Disclosure document drafting

Step-by-step instructions are in prompts/(within this package). Invention / utility model / design arethree directories within the package, not three triggerable skills.

Step File
Step 1 prompts/intake.md
Step 2 prompts/project_scan.md
Step 3–4 prompts/invention/ · utility_model/ · design/ point mining
form filling / line drawings prompts/fill_*、image_gen.md、*_lineart_*.md; design view conventions references/design_view_cnipa.md
Step 5 prompts/prior_art_search.md(lightweight novelty search, one term per page)
Step 6 prompts/disclosure_preview.md
Step 7 corresponding type disclosure_builder.md + template_reference.md
Step 8 prompts/disclosure_self_check.md
iteration iteration_context.md / merger.md / correction_handler.md

Novelty search tool:tools/crawl/cnipa_epub_search.py. When installed repo-wide, the path is skills/patent-disclosure/tools/crawl/cnipa_epub_search.py. Bibliographic search is not in this package,forbiddento be invoked as a novelty search engine.
After the disclosure is delivereddo notautomatically proceed to the application documents; only after the user names it and provides this directory, then proceed to skills/patent-application/SKILL.md。
--type consistent with intake; two-stage: keywords → EPUB_CLASS_HINT / IPC·LOC → --class; if fewer than 4 items, supplement the first round with the same classification number.

For line drawings, CAD, formulas, and Word figure export, use this package's tools/(browser.py、mermaid_render.py、md_to_docx.py , etc.).