Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Sweep your folders and downloads for tax documents and build a CPA-ready package -- checklist by form type, everything renamed and organized, a missing-document chase list, and flagged items worth asking your accountant about. Organization, not tax advice.
.claude/skills/onewave-ai-cowork-tax-prep-organizer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 163% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 80% | 0% |
Do the part of tax season that actually eats the weekend: finding, identifying, and organizing the documents. Input: one or more folders (Downloads, a documents dump, last year's tax folder for reference). Output: an organized package a CPA can work from and a precise list of what is still missing. This skill organizes; it does not compute taxes or give tax advice.
tax-2025/income/, deductions/, investments/, business/, prior-year/, with names like 1099-nec-acme-corp-2025.pdf. Write INDEX.md mapping checklist to files.As a monthly Cowork scheduled task during January-April: re-sweep Downloads and the documents folder for newly arrived forms, file them into the package, and update the chase list -- so tax weekend becomes tax hour.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 9,910 | 7,523 | -24% | 1 | 1 | 0% | 1,712 | 1,359 | -21% | 0 | 0 | — |
case-02 | fail→fail | 10,238 | 10,461 | +2% | 1 | 1 | 0% | 1,438 | 2,336 | +62% | 0 | 0 | — |
case-03 | fail→fail | 18,714 | 8,095 | -57% | 1 | 1 | 0% | 3,176 | 2,083 | -34% | 0 | 0 | — |
case-04 | fail→pass | 5,528 | 3,267 | -41% | 1 | 1 | 0% | 932 | 1,245 | +34% | 0 | 0 | — |
case-05 | fail→pass | 12,990 | 7,457 | -43% | 1 | 1 | 0% | 2,351 | 1,990 | -15% | 0 | 0 | — |
case-06 | fail→pass | 3,726 | 4,814 | +29% | 1 | 1 | 0% | 566 | 1,489 | +163% | 0 | 0 | — |
case-07 | fail→pass | 9,468 | 7,056 | -25% | 1 | 1 | 0% | 1,717 | 1,804 | +5% | 0 | 0 | — |
case-08 | pass→pass | 3,183 | 5,826 | +83% | 1 | 1 | 0% | 495 | 1,715 | +246% | 0 | 0 | — |
case-09 | pass→pass | 8,546 | 9,349 | +9% | 1 | 1 | 0% | 1,393 | 2,204 | +58% | 0 | 0 | — |
case-10 | fail→pass | 8,058 | 9,619 | +19% | 1 | 1 | 0% | 1,288 | 2,324 | +80% | 0 | 0 | — |
case-11 | fail→fail | 15,061 | 12,564 | -17% | 1 | 1 | 0% | 2,616 | 2,800 | +7% | 0 | 0 | — |
case-12 | pass→pass | 12,906 | 10,726 | -17% | 1 | 1 | 0% | 1,970 | 2,361 | +20% | 0 | 0 | — |
case-13 | fail→pass | 7,339 | 13,322 | +82% | 1 | 1 | 0% | 1,104 | 2,914 | +164% | 0 | 0 | — |
case-14 | pass→pass | 6,383 | 4,591 | -28% | 1 | 1 | 0% | 1,108 | 1,410 | +27% | 0 | 0 | — |
case-15 | pass→pass | 13,500 | 10,324 | -24% | 1 | 1 | 0% | 2,246 | 2,357 | +5% | 0 | 0 | — |
case-16 | fail→pass | 9,176 | 12,590 | +37% | 1 | 1 | 0% | 1,583 | 2,804 | +77% | 0 | 0 | — |
case-17 | fail→pass | 17,444 | 9,965 | -43% | 1 | 1 | 0% | 2,671 | 2,104 | -21% | 0 | 0 | — |
case-18 | fail→fail | 6,936 | 12,468 | +80% | 1 | 1 | 0% | 1,038 | 2,690 | +159% | 0 | 0 | — |
case-19 | pass→pass | 9,601 | 7,351 | -23% | 1 | 1 | 0% | 1,548 | 1,742 | +13% | 0 | 0 | — |
case-20 | pass→pass | 7,357 | 4,874 | -34% | 1 | 1 | 0% | 1,301 | 1,592 | +22% | 0 | 0 | — |
case-21 | fail→pass | 9,885 | 7,078 | -28% | 1 | 1 | 0% | 1,527 | 1,843 | +21% | 0 | 0 | — |
case-22 | fail→pass | 10,426 | 8,000 | -23% | 1 | 1 | 0% | 1,770 | 2,073 | +17% | 0 | 0 | — |
case-23 | pass→pass | 12,569 | 9,380 | -25% | 1 | 1 | 0% | 1,990 | 2,276 | +14% | 0 | 0 | — |
case-24 | pass→pass | 4,531 | 10,917 | +141% | 1 | 1 | 0% | 663 | 2,525 | +281% | 0 | 0 | — |
case-25 | fail→fail | 9,912 | 8,801 | -11% | 1 | 1 | 0% | 1,648 | 2,279 | +38% | 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. 25 cases were attempted. The headline lift of +40 percentage points is the difference between those two pass rates over the 25 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.