Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Compile patent application into jurisdiction-specific filing format. Use when user says "格式转换", "jurisdiction format", "国家格式", "compile patent", or wants formatted patent documents for CN/US/EP filing.
.claude/skills/thomasmoreai-jurisdiction-format/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 1433% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 2134% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 57% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 58% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 26% | 0% |
Compile the patent application into filing-ready format based on: $ARGUMENTS
Analogous to /paper-compile but for patent document formatting instead of LaTeX.
JURISDICTION = "auto" — From pipeline or args: CN, US, EP, ALLPATENT_TYPE = "invention" — invention (发明专利) or utility_model (实用新型, CN only)OUTPUT_FORMAT = "markdown" — markdown (for review) or docx (for filing, requires python-docx)OUTPUT_DIR = "patent/output/" — Base output directorypatent/CLAIMS.md — drafted claimspatent/specification/ — all specification sections (title, technical_field, background, summary, drawings_description, detailed_description, abstract)patent/figures/ — figure descriptions and numeral indexpatent/INVENTION_DISCLOSURE.md — for metadataLoad ../shared-references/patent-format-cn.md for CNIPA document structure and formatting rules. Load ../shared-references/patent-format-us.md for USPTO document structure. Load ../shared-references/patent-format-ep.md for EPO document structure.
From $ARGUMENTS or constant:
CN -> Generate CNIPA format onlyUS -> Generate USPTO format onlyEP -> Generate EPO format onlyALL -> Generate all three formatsOutput to patent/output/CN/:
patent/CLAIMS.mdCombine all specification sections in CNIPA order:
If OUTPUT_FORMAT = "markdown":
If OUTPUT_FORMAT = "docx":
Output to patent/output/US/:
Combine all sections in USPTO order:
Format drawings references as "FIG. 1" (not "Figure 1").
Generate a skeleton ADS with:
Output to patent/output/EP/:
Combine all sections in EPO Rule 42 order:
Format drawings references as "FIG. 1" or "Figure 1".
Verify across all generated formats:
Write patent/output/OUTPUT_SUMMARY.md:
markdown## Patent Filing Documents ### Generated Files #### CN (CNIPA) | File | Description | Status | |------|-------------|--------| | 权利要求书.md | Claims in CN format | Complete | | 说明书.md | Description in CN format | Complete | | 说明书摘要.md | Abstract (CN) | Complete | #### US (USPTO) | File | Description | Status | |------|-------------|--------| | claims.md | Claims in US format | Complete | | specification.md | Description in US format | Complete | | abstract.md | Abstract (US) | Complete | | ads_template.md | Application Data Sheet skeleton | Complete | #### EP (EPO) | File | Description | Status | |------|-------------|--------| | claims.md | Claims in EP format | Complete | | description.md | Description in EP format | Complete | | abstract.md | Abstract (EP) | Complete | ### Consistency Check - [ ] All claims present in all formats - [ ] Reference numerals consistent - [ ] Language correct per jurisdiction
OUTPUT_FORMAT = "docx", check that python-docx is available; if not, fall back to markdown.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 6,530 | 21,256 | +226% | 1 | 1 | 0% | 397 | 6,085 | +1433% | 0 | 0 | — |
case-02 | fail→fail | 4,699 | 2,955 | -37% | 1 | 1 | 0% | 266 | 2,039 | +667% | 0 | 0 | — |
case-03 | fail→pass | 4,198 | 20,459 | +387% | 1 | 1 | 0% | 228 | 5,093 | +2134% | 0 | 0 | — |
case-04 | fail→fail | 23,803 | 5,666 | -76% | 1 | 1 | 0% | 3,452 | 1,984 | -43% | 0 | 0 | — |
case-05 | fail→pass | 11,094 | 10,096 | -9% | 1 | 1 | 0% | 2,097 | 3,300 | +57% | 0 | 0 | — |
case-06 | pass→pass | 6,542 | 4,680 | -28% | 1 | 1 | 0% | 1,043 | 2,474 | +137% | 0 | 0 | — |
case-07 | fail→fail | 4,817 | 3,799 | -21% | 1 | 1 | 0% | 234 | 1,788 | +664% | 0 | 0 | — |
case-08 | pass→pass | 14,069 | 10,058 | -29% | 1 | 1 | 0% | 2,342 | 3,409 | +46% | 0 | 0 | — |
case-09 | fail→pass | 9,131 | 6,587 | -28% | 1 | 1 | 0% | 1,708 | 2,698 | +58% | 0 | 0 | — |
case-10 | fail→pass | 15,313 | 10,186 | -33% | 1 | 1 | 0% | 2,807 | 3,535 | +26% | 0 | 0 | — |
case-11 | fail→fail | 13,174 | 6,172 | -53% | 1 | 1 | 0% | 1,991 | 2,743 | +38% | 0 | 0 | — |
case-12 | pass→pass | 12,442 | 4,141 | -67% | 1 | 1 | 0% | 2,047 | 2,345 | +15% | 0 | 0 | — |
case-13 | fail→pass | 11,693 | 3,891 | -67% | 1 | 1 | 0% | 1,606 | 2,446 | +52% | 0 | 0 | — |
case-14 | fail→pass | 11,492 | 4,375 | -62% | 1 | 1 | 0% | 2,093 | 2,448 | +17% | 0 | 0 | — |
case-15 | pass→pass | 13,383 | 5,553 | -59% | 1 | 1 | 0% | 2,469 | 2,440 | -1% | 0 | 0 | — |
case-16 | pass→pass | 14,102 | 13,385 | -5% | 1 | 1 | 0% | 2,441 | 3,518 | +44% | 0 | 0 | — |
case-17 | fail→pass | 12,844 | 9,563 | -26% | 1 | 1 | 0% | 2,166 | 3,326 | +54% | 0 | 0 | — |
case-18 | fail→pass | 8,256 | 5,942 | -28% | 1 | 1 | 0% | 1,477 | 2,580 | +75% | 0 | 0 | — |
case-19 | pass→pass | 5,825 | 1,881 | -68% | 1 | 1 | 0% | 926 | 1,903 | +106% | 0 | 0 | — |
case-20 | fail→fail | 6,788 | 12,897 | +90% | 1 | 1 | 0% | 1,071 | 3,818 | +256% | 0 | 0 | — |
case-21 | pass→fail | 5,324 | 8,020 | +51% | 1 | 1 | 0% | 850 | 3,050 | +259% | 0 | 0 | — |
case-22 | fail→fail | 4,003 | 11,897 | +197% | 1 | 1 | 0% | 650 | 3,884 | +498% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 17 counted toward the lift figure. The other 5 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +36 percentage points is the difference between those two pass rates over the 17 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.