Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Audit your calendar with Cowork's calendar tools -- measure meeting load and fragmentation, identify which recurring meetings earn their slot, propose consolidations and focus blocks, and draft the diplomatic messages that reclaim your week.
.claude/skills/onewave-ai-cowork-calendar-defrag/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 48% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -15% | 0% |
Treat the calendar like a disk that needs defragmenting: measure what is actually there, identify waste, consolidate, and carve out contiguous space for real work. Uses the connected calendar tools (Microsoft 365 / Google Calendar) to read events; every change is proposed, and the human approves before anything is created, moved, or declined.
calendar-defrag-report.md with before/after metrics: meeting hours reclaimed, longest block gained.As a monthly Cowork scheduled task: re-measure, compare against the last report, flag regression (creeping recurrings, eroded focus blocks), and propose the next round of cuts.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 17,113 | 19,039 | +11% | 1 | 1 | 0% | 2,623 | 3,877 | +48% | 0 | 0 | — |
case-02 | fail→pass | 19,319 | 13,859 | -28% | 1 | 1 | 0% | 2,721 | 2,434 | -11% | 0 | 0 | — |
case-03 | fail→pass | 13,598 | 3,520 | -74% | 1 | 1 | 0% | 1,841 | 1,127 | -39% | 0 | 0 | — |
case-15 | pass→pass | 13,131 | 8,240 | -37% | 1 | 1 | 0% | 1,788 | 1,996 | +12% | 0 | 0 | — |
case-04 | pass→pass | 11,960 | 2,029 | -83% | 1 | 1 | 0% | 1,565 | 962 | -39% | 0 | 0 | — |
case-05 | fail→pass | 11,724 | 3,197 | -73% | 1 | 1 | 0% | 1,657 | 1,192 | -28% | 0 | 0 | — |
case-06 | fail→pass | 10,665 | 4,122 | -61% | 1 | 1 | 0% | 1,509 | 1,278 | -15% | 0 | 0 | — |
case-07 | pass→pass | 12,734 | 10,253 | -19% | 1 | 1 | 0% | 1,861 | 2,220 | +19% | 0 | 0 | — |
case-08 | pass→pass | 10,855 | 6,945 | -36% | 1 | 1 | 0% | 1,648 | 1,682 | +2% | 0 | 0 | — |
case-09 | pass→pass | 12,438 | 8,973 | -28% | 1 | 1 | 0% | 2,030 | 2,067 | +2% | 0 | 0 | — |
case-10 | pass→pass | 10,855 | 7,999 | -26% | 1 | 1 | 0% | 1,855 | 2,023 | +9% | 0 | 0 | — |
case-11 | pass→pass | 12,390 | 5,936 | -52% | 1 | 1 | 0% | 1,591 | 1,632 | +3% | 0 | 0 | — |
case-12 | fail→fail | 8,878 | 1,775 | -80% | 1 | 1 | 0% | 1,250 | 976 | -22% | 0 | 0 | — |
case-13 | pass→pass | 10,850 | 2,500 | -77% | 1 | 1 | 0% | 1,532 | 1,087 | -29% | 0 | 0 | — |
case-14 | fail→pass | 12,053 | 3,495 | -71% | 1 | 1 | 0% | 1,879 | 1,284 | -32% | 0 | 0 | — |
case-16 | pass→pass | 12,703 | 3,498 | -72% | 1 | 1 | 0% | 1,932 | 1,208 | -37% | 0 | 0 | — |
case-17 | pass→pass | 10,440 | 9,455 | -9% | 1 | 1 | 0% | 1,617 | 2,057 | +27% | 0 | 0 | — |
case-18 | fail→pass | 12,328 | 9,005 | -27% | 1 | 1 | 0% | 1,886 | 2,211 | +17% | 0 | 0 | — |
case-19 | fail→fail | 6,919 | 6,142 | -11% | 1 | 1 | 0% | 1,028 | 1,436 | +40% | 0 | 0 | — |
case-20 | pass→pass | 5,639 | 3,371 | -40% | 1 | 1 | 0% | 739 | 1,208 | +63% | 0 | 0 | — |
case-21 | pass→pass | 8,243 | 8,738 | +6% | 1 | 1 | 0% | 1,561 | 2,194 | +41% | 0 | 0 | — |
case-22 | pass→pass | 9,043 | 13,993 | +55% | 1 | 1 | 0% | 1,463 | 1,709 | +17% | 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. The headline lift of +32 percentage points is the difference between those two pass rates over the 22 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.