Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Write detailed embodiment descriptions for patent specifications. Use when user says "撰写实施例", "write embodiment", "实施例描述", "detailed description", or wants to describe how to practice an invention.
.claude/skills/wanshuiyin-embodiment-description/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 319% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 77% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 93% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 78% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 110% | 0% |
Write detailed embodiments for: $ARGUMENTS
Embodiments describe HOW to make and use the invention -- they are the patent equivalent of experiment sections, but describe the invention rather than evaluating it empirically.
MIN_EMBODIMENTS = 1 — At least one complete embodiment requiredMAX_EMBODIMENTS = 3 — Practical limit; more embodiments strengthen enablementEMBODIMENT_STYLE = detailed — detailed (full working example) or outline (sketch)REFERENCE_NUMERAL_PREFIX = 100 — Starting reference numeral for first figure's componentspatent/INVENTION_DISCLOSURE.md — invention decomposition (core/supporting/optional features)patent/CLAIMS.md — drafted claims that the embodiments must supportpatent/figures/numeral_index.md if it exists (from /figure-description)For each claim category (method, system, etc.), plan at least one embodiment:
| Embodiment | Covers Claims | Type | Key Variations | |-----------|--------------|------|----------------| | 1 | Claims 1, X | Best mode / preferred | primary implementation] | | 2 | Claims 2, 3 | Alternative | different parameters/materials] | | 3 | Claims 4, 5 | Additional alternative | different configuration] |
For each embodiment, write a detailed description following this structure:
Opening paragraph: "In one embodiment, invention summary with reference to what is being described]."
Component/step-by-step description:
For method embodiments:
For system/apparatus embodiments:
Variations and alternatives:
These variations are critical -- they support broader claim interpretation.
Ensure consistent reference numeral usage:
Format:
For each claim element, verify it appears in at least one embodiment:
| Claim Element | Embodiment | Reference Numeral | Description Paragraph | |---------------|-----------|-------------------|----------------------| | element] | which] | numeral] | paragraph reference] |
If any claim element lacks embodiment support, add the necessary description.
For method/software inventions, include:
Example:
In one embodiment, the method comprises the following steps:
At step 202, the processor 102 receives input data from the input device 108.
At step 204, the processor 102 extracts feature vectors from the input data using a convolutional neural network.
At step 206, the processor 102 applies the attention mechanism 110 to the feature vectors...Embodiment sections are written to patent/specification/detailed_description.md (or appended to the specification structure).
Each embodiment section should be self-contained but cross-reference other embodiments when describing alternatives.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 26,554 | 4,075 | -85% | 1 | 1 | 0% | 5,046 | 1,751 | -65% | 0 | 0 | — |
case-02 | fail→pass | 10,547 | 33,050 | +213% | 1 | 1 | 0% | 1,836 | 7,697 | +319% | 0 | 0 | — |
case-03 | fail→fail | 13,771 | 27,628 | +101% | 1 | 1 | 0% | 2,730 | 6,960 | +155% | 0 | 0 | — |
case-04 | fail→pass | 19,521 | 24,936 | +28% | 1 | 1 | 0% | 3,604 | 6,365 | +77% | 0 | 0 | — |
case-05 | fail→pass | 14,680 | 17,281 | +18% | 1 | 1 | 0% | 2,631 | 5,075 | +93% | 0 | 0 | — |
case-06 | pass→pass | 14,951 | 16,677 | +12% | 1 | 1 | 0% | 2,878 | 4,948 | +72% | 0 | 0 | — |
case-21 | fail→pass | 14,585 | 18,811 | +29% | 1 | 1 | 0% | 2,934 | 5,222 | +78% | 0 | 0 | — |
case-07 | pass→fail | 25,065 | 24,716 | -1% | 1 | 1 | 0% | 4,848 | 6,797 | +40% | 0 | 0 | — |
case-08 | pass→pass | 18,954 | 25,058 | +32% | 1 | 1 | 0% | 3,300 | 6,370 | +93% | 0 | 0 | — |
case-09 | pass→pass | 13,498 | 30,507 | +126% | 1 | 1 | 0% | 2,399 | 7,149 | +198% | 0 | 0 | — |
case-10 | pass→fail | 12,272 | 30,520 | +149% | 1 | 1 | 0% | 2,212 | 7,653 | +246% | 0 | 0 | — |
case-11 | pass→pass | 30,476 | 24,955 | -18% | 1 | 1 | 0% | 6,163 | 7,658 | +24% | 0 | 0 | — |
case-12 | fail→pass | 17,890 | 22,247 | +24% | 1 | 1 | 0% | 3,066 | 6,439 | +110% | 0 | 0 | — |
case-13 | pass→pass | 18,399 | 26,926 | +46% | 1 | 1 | 0% | 3,686 | 7,652 | +108% | 0 | 0 | — |
case-14 | pass→pass | 11,947 | 19,543 | +64% | 1 | 1 | 0% | 2,254 | 5,761 | +156% | 0 | 0 | — |
case-19 | fail→pass | 17,584 | 19,645 | +12% | 1 | 1 | 0% | 3,364 | 5,556 | +65% | 0 | 0 | — |
case-15 | pass→pass | 16,540 | 26,654 | +61% | 1 | 1 | 0% | 3,022 | 7,084 | +134% | 0 | 0 | — |
case-16 | fail→pass | 16,137 | 27,641 | +71% | 1 | 1 | 0% | 2,982 | 7,443 | +150% | 0 | 0 | — |
case-17 | pass→pass | 14,912 | 19,543 | +31% | 1 | 1 | 0% | 2,848 | 5,594 | +96% | 0 | 0 | — |
case-18 | fail→fail | 5,780 | 5,413 | -6% | 1 | 1 | 0% | 1,079 | 2,551 | +136% | 0 | 0 | — |
case-20 | pass→pass | 17,150 | 28,363 | +65% | 1 | 1 | 0% | 3,148 | 7,655 | +143% | 0 | 0 | — |
case-22 | fail→pass | 15,100 | 7,038 | -53% | 1 | 1 | 0% | 2,528 | 2,649 | +5% | 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 21 counted toward the lift figure. The other 1 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 +27 percentage points is the difference between those two pass rates over the 21 comparable cases. 3 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.