Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Research a topic from the past X days using web search. Specify the number of days or let it be inferred from your prompt.
.claude/skills/cowork-os-research-last-days/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✓→✓ | ＝ Same ✓ | -26% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -20% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | -35% | 0% |
| case-01 | ✗→✗ | ＝ Same ✗ | -64% | 0% |
| case-02 | ✗→✗ | ＝ Same ✗ | -69% | 0% |
Research a topic from the past X days using web search. Specify the number of days or let it be inferred from your prompt.
| Name | Type | Required | Description | |---|---|---|---| | topic | string | Yes | The topic to research (e.g., 'AI agents', 'prompting techniques', 'Apple announcements') | | days | number | No | Number of days to look back (e.g., 1, 7, 14, 30, 90). Defaults to 7. | | tool | select | No | Target AI tool for the prompts (optional context) |
../research-last-days.json.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 42,731 | 9,075 | -79% | 1 | 1 | 0% | 2,708 | 987 | -64% | 0 | 0 | — |
case-02 | fail→fail | 17,067 | 20,753 | +22% | 1 | 1 | 0% | 2,406 | 735 | -69% | 0 | 0 | — |
case-03 | fail→fail | 15,521 | 16,509 | +6% | 1 | 1 | 0% | 1,893 | 697 | -63% | 0 | 0 | — |
case-04 | pass→pass | 17,899 | 10,860 | -39% | 1 | 1 | 0% | 2,873 | 2,135 | -26% | 0 | 0 | — |
case-05 | pass→pass | 17,768 | 11,460 | -36% | 1 | 1 | 0% | 2,853 | 2,283 | -20% | 0 | 0 | — |
case-06 | pass→pass | 17,709 | 9,413 | -47% | 1 | 1 | 0% | 2,845 | 1,854 | -35% | 0 | 0 | — |
case-07 | fail→fail | 13,939 | 7,348 | -47% | 1 | 1 | 0% | 2,171 | 784 | -64% | 0 | 0 | — |
case-08 | fail→fail | 8,306 | 6,788 | -18% | 1 | 1 | 0% | 1,310 | 784 | -40% | 0 | 0 | — |
case-09 | fail→fail | 15,641 | 24,587 | +57% | 1 | 1 | 0% | 2,107 | 835 | -60% | 0 | 0 | — |
case-10 | fail→fail | 3,550 | 6,077 | +71% | 1 | 1 | 0% | 562 | 740 | +32% | 0 | 0 | — |
case-11 | fail→fail | 30,444 | 17,312 | -43% | 1 | 1 | 0% | 398 | 769 | +93% | 0 | 0 | — |
case-12 | fail→fail | 12,533 | 25,154 | +101% | 1 | 1 | 0% | 1,875 | 3,311 | +77% | 0 | 0 | — |
case-13 | fail→fail | 2,795 | 5,024 | +80% | 1 | 1 | 0% | 422 | 666 | +58% | 0 | 0 | — |
case-14 | fail→fail | 13,140 | 7,905 | -40% | 1 | 1 | 0% | 2,323 | 944 | -59% | 0 | 0 | — |
case-15 | fail→fail | 11,872 | 6,928 | -42% | 1 | 1 | 0% | 1,614 | 813 | -50% | 0 | 0 | — |
case-16 | fail→fail | 14,278 | 12,601 | -12% | 1 | 1 | 0% | 2,162 | 698 | -68% | 0 | 0 | — |
case-17 | fail→fail | 18,134 | 5,660 | -69% | 1 | 1 | 0% | 3,143 | 804 | -74% | 0 | 0 | — |
case-18 | fail→fail | 5,741 | 20,865 | +263% | 1 | 1 | 0% | 917 | 2,236 | +144% | 0 | 0 | — |
case-19 | fail→fail | 20,411 | 12,416 | -39% | 1 | 1 | 0% | 1,924 | 2,262 | +18% | 0 | 0 | — |
case-20 | fail→fail | 10,577 | 21,795 | +106% | 1 | 1 | 0% | 1,706 | 756 | -56% | 0 | 0 | — |
case-21 | fail→fail | 17,607 | 7,951 | -55% | 1 | 1 | 0% | 2,764 | 772 | -72% | 0 | 0 | — |
case-22 | fail→fail | 6,693 | 8,775 | +31% | 1 | 1 | 0% | 983 | 780 | -21% | 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 5 counted toward the lift figure. The other 17 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 0 percentage points is the difference between those two pass rates over the 5 comparable cases. 12 cases got worse with the skill loaded, and they are included in that figure.
Other measured skills in the registry, with their headline benchmark lift.