Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Turn a process walkthrough -- meeting transcript, screen-recording transcript, rough notes, or a folder of scattered how-to docs -- into a clean standard operating procedure: numbered steps, roles, decision points, exceptions, and a consistent template across your whole SOP library.
.claude/skills/onewave-ai-cowork-sop-writer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | 153% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 61% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 98% | 0% |
Write procedures the way a good operations manager does: capture how the work is actually done, make the implicit explicit, and write for the new hire on their worst day -- not for the expert who already knows. Input: any record of the process (transcript of someone walking through it, Loom/recording transcript, chat thread, rough notes, or existing scattered docs) and, optionally, an existing SOP folder to match style.
[TO CONFIRM] inline rather than guessing.[SCREENSHOT: what it should show] placeholders where a visual would prevent an error.sop-[team]-[process-name].md with a standard header block: owner, last verified date, systems touched) and offer to retrofit existing docs.[TO CONFIRM] beats plausible fiction that a new hire will execute literally.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 15,327 | 8,362 | -45% | 1 | 1 | 0% | 2,555 | 1,683 | -34% | 0 | 0 | — |
case-02 | pass→fail | 10,816 | 12,531 | +16% | 1 | 1 | 0% | 1,788 | 2,704 | +51% | 0 | 0 | — |
case-03 | pass→pass | 25,306 | 20,556 | -19% | 1 | 1 | 0% | 3,676 | 3,894 | +6% | 0 | 0 | — |
case-04 | pass→pass | 17,106 | 20,308 | +19% | 1 | 1 | 0% | 3,389 | 4,573 | +35% | 0 | 0 | — |
case-05 | pass→pass | 7,262 | 15,222 | +110% | 1 | 1 | 0% | 1,383 | 3,292 | +138% | 0 | 0 | — |
case-06 | pass→pass | 7,264 | 8,685 | +20% | 1 | 1 | 0% | 1,159 | 2,121 | +83% | 0 | 0 | — |
case-07 | fail→fail | 13,382 | 8,689 | -35% | 1 | 1 | 0% | 2,094 | 1,971 | -6% | 0 | 0 | — |
case-13 | pass→pass | 9,514 | 9,584 | +1% | 1 | 1 | 0% | 1,668 | 2,271 | +36% | 0 | 0 | — |
case-08 | fail→pass | 4,702 | 8,334 | +77% | 1 | 1 | 0% | 791 | 2,003 | +153% | 0 | 0 | — |
case-09 | fail→pass | 11,360 | 13,348 | +18% | 1 | 1 | 0% | 1,809 | 2,912 | +61% | 0 | 0 | — |
case-10 | fail→pass | 19,214 | 16,269 | -15% | 1 | 1 | 0% | 2,965 | 3,236 | +9% | 0 | 0 | — |
case-11 | pass→pass | 8,692 | 12,266 | +41% | 1 | 1 | 0% | 1,447 | 2,677 | +85% | 0 | 0 | — |
case-12 | fail→pass | 17,336 | 20,294 | +17% | 1 | 1 | 0% | 2,877 | 3,884 | +35% | 0 | 0 | — |
case-14 | fail→pass | 4,412 | 4,766 | +8% | 1 | 1 | 0% | 682 | 1,349 | +98% | 0 | 0 | — |
case-15 | pass→pass | 12,768 | 16,639 | +30% | 1 | 1 | 0% | 1,943 | 3,094 | +59% | 0 | 0 | — |
case-16 | pass→pass | 7,039 | 7,913 | +12% | 1 | 1 | 0% | 1,175 | 2,006 | +71% | 0 | 0 | — |
case-17 | pass→pass | 10,714 | 17,238 | +61% | 1 | 1 | 0% | 1,798 | 3,442 | +91% | 0 | 0 | — |
case-18 | pass→pass | 18,975 | 18,512 | -2% | 1 | 1 | 0% | 2,710 | 3,564 | +32% | 0 | 0 | — |
case-19 | pass→pass | 8,425 | 10,726 | +27% | 1 | 1 | 0% | 1,381 | 2,489 | +80% | 0 | 0 | — |
case-20 | pass→pass | 14,028 | 12,978 | -7% | 1 | 1 | 0% | 2,071 | 2,488 | +20% | 0 | 0 | — |
case-21 | pass→pass | 12,493 | 11,225 | -10% | 1 | 1 | 0% | 1,986 | 2,462 | +24% | 0 | 0 | — |
case-22 | fail→fail | 3,902 | 5,540 | +42% | 1 | 1 | 0% | 602 | 1,522 | +153% | 0 | 0 | — |
case-23 | pass→pass | 2,832 | 7,125 | +152% | 1 | 1 | 0% | 478 | 1,915 | +301% | 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. 23 cases were attempted. The headline lift of +17 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is 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.