Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Comprehensive document creation, editing, and analysis with support for tracked changes, comments, formatting preservation, and text extraction. When Claude needs to work with professional documents (.docx files) for: (1) Creating new documents, (2) Modifying or editing content, (3) Working with tracked changes, (4) Adding comments, or any other document tasks
.claude/skills/dokhacgiakhoa-docx/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 62% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 94% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | -4% | 0% |
A user may ask you to create, edit, or analyze the contents of a .docx file. A .docx file is essentially a ZIP archive containing XML files and other resources that you can read or edit. You have different tools and workflows available for different tasks.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 3,935 | 23,743 | +503% | 1 | 1 | 0% | 574 | 575 | +0% | 0 | 0 | — |
case-02 | pass→pass | 12,366 | 15,309 | +24% | 1 | 1 | 0% | 1,875 | 3,641 | +94% | 0 | 0 | — |
case-03 | pass→pass | 16,469 | 13,016 | -21% | 1 | 1 | 0% | 2,973 | 2,859 | -4% | 0 | 0 | — |
case-04 | pass→pass | 12,210 | 9,280 | -24% | 1 | 1 | 0% | 2,027 | 1,704 | -16% | 0 | 0 | — |
case-05 | fail→pass | 14,316 | 13,365 | -7% | 1 | 1 | 0% | 2,048 | 2,552 | +25% | 0 | 0 | — |
case-06 | fail→pass | 8,335 | 10,449 | +25% | 1 | 1 | 0% | 1,270 | 1,948 | +53% | 0 | 0 | — |
case-07 | pass→pass | 9,168 | 5,694 | -38% | 1 | 1 | 0% | 1,302 | 1,274 | -2% | 0 | 0 | — |
case-08 | pass→pass | 13,318 | 13,683 | +3% | 1 | 1 | 0% | 2,146 | 2,545 | +19% | 0 | 0 | — |
case-09 | pass→pass | 13,897 | 12,629 | -9% | 1 | 1 | 0% | 2,317 | 2,388 | +3% | 0 | 0 | — |
case-10 | pass→pass | 11,720 | 8,244 | -30% | 1 | 1 | 0% | 1,700 | 1,403 | -17% | 0 | 0 | — |
case-11 | pass→pass | 8,889 | 5,264 | -41% | 1 | 1 | 0% | 1,342 | 1,260 | -6% | 0 | 0 | — |
case-12 | fail→fail | 15,027 | 15,010 | -0% | 1 | 1 | 0% | 2,596 | 2,887 | +11% | 0 | 0 | — |
case-13 | pass→pass | 14,945 | 14,785 | -1% | 1 | 1 | 0% | 2,541 | 2,695 | +6% | 0 | 0 | — |
case-14 | pass→pass | 16,624 | 11,184 | -33% | 1 | 1 | 0% | 2,580 | 2,180 | -16% | 0 | 0 | — |
case-15 | pass→pass | 11,205 | 11,991 | +7% | 1 | 1 | 0% | 2,261 | 2,555 | +13% | 0 | 0 | — |
case-16 | pass→pass | 13,668 | 13,268 | -3% | 1 | 1 | 0% | 2,596 | 2,906 | +12% | 0 | 0 | — |
case-17 | fail→pass | 8,163 | 12,340 | +51% | 1 | 1 | 0% | 1,440 | 2,331 | +62% | 0 | 0 | — |
case-18 | pass→pass | 9,893 | 7,218 | -27% | 1 | 1 | 0% | 1,750 | 1,639 | -6% | 0 | 0 | — |
case-19 | pass→pass | 8,614 | 5,971 | -31% | 1 | 1 | 0% | 1,459 | 1,278 | -12% | 0 | 0 | — |
case-20 | pass→pass | 13,736 | 15,229 | +11% | 1 | 1 | 0% | 2,303 | 3,073 | +33% | 0 | 0 | — |
case-21 | pass→pass | 17,131 | 14,514 | -15% | 1 | 1 | 0% | 2,857 | 2,822 | -1% | 0 | 0 | — |
case-22 | pass→pass | 18,131 | 10,190 | -44% | 1 | 1 | 0% | 2,607 | 2,022 | -22% | 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 +14 percentage points is the difference between those two pass rates over the 21 comparable cases.
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.