Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Write job posts that make the right person apply -- real role clarity, honest selling points, salary range handling, and requirements pruned to what actually predicts success. Small businesses lose great candidates to bad posts long before the interview. Feeds cowork-hiring-screener and hiring-scorecard downstream.
.claude/skills/onewave-ai-job-post-writer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 52% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 65% | 0% |
A job post is an ad competing against every employer in the market, and most read like compliance documents. Write the one the right candidate stops scrolling for -- honest about the work, specific about the money, and stripped of the requirements that filter out people who would have been great. Input: what the role actually does day-to-day (interviewed out of the owner if needed) and what the business can truly offer.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 13,116 | 15,129 | +15% | 1 | 1 | 0% | 1,994 | 2,883 | +45% | 0 | 0 | — |
case-02 | fail→fail | 7,216 | 5,495 | -24% | 1 | 1 | 0% | 1,179 | 1,601 | +36% | 0 | 0 | — |
case-03 | pass→pass | 13,284 | 14,985 | +13% | 1 | 1 | 0% | 1,936 | 2,665 | +38% | 0 | 0 | — |
case-04 | pass→pass | 16,874 | 16,722 | -1% | 1 | 1 | 0% | 2,533 | 2,915 | +15% | 0 | 0 | — |
case-05 | fail→fail | 13,153 | 14,347 | +9% | 1 | 1 | 0% | 1,852 | 2,589 | +40% | 0 | 0 | — |
case-06 | pass→fail | 12,750 | 16,491 | +29% | 1 | 1 | 0% | 1,843 | 2,945 | +60% | 0 | 0 | — |
case-07 | pass→fail | 7,378 | 9,098 | +23% | 1 | 1 | 0% | 1,218 | 2,113 | +73% | 0 | 0 | — |
case-08 | pass→pass | 13,797 | 15,337 | +11% | 1 | 1 | 0% | 2,027 | 3,055 | +51% | 0 | 0 | — |
case-09 | fail→pass | 12,378 | 9,958 | -20% | 1 | 1 | 0% | 1,898 | 2,214 | +17% | 0 | 0 | — |
case-10 | fail→fail | 12,842 | 8,786 | -32% | 1 | 1 | 0% | 1,917 | 1,982 | +3% | 0 | 0 | — |
case-11 | fail→pass | 8,568 | 9,269 | +8% | 1 | 1 | 0% | 1,357 | 2,059 | +52% | 0 | 0 | — |
case-12 | pass→fail | 10,156 | 7,418 | -27% | 1 | 1 | 0% | 1,580 | 1,859 | +18% | 0 | 0 | — |
case-13 | fail→pass | 9,090 | 9,186 | +1% | 1 | 1 | 0% | 1,723 | 2,316 | +34% | 0 | 0 | — |
case-14 | fail→pass | 9,268 | 11,103 | +20% | 1 | 1 | 0% | 1,732 | 2,643 | +53% | 0 | 0 | — |
case-15 | pass→pass | 12,960 | 8,123 | -37% | 1 | 1 | 0% | 2,114 | 2,069 | -2% | 0 | 0 | — |
case-16 | fail→fail | 9,457 | 9,349 | -1% | 1 | 1 | 0% | 1,636 | 2,142 | +31% | 0 | 0 | — |
case-17 | pass→pass | 17,071 | 14,224 | -17% | 1 | 1 | 0% | 2,273 | 2,724 | +20% | 0 | 0 | — |
case-18 | fail→pass | 7,684 | 10,833 | +41% | 1 | 1 | 0% | 1,448 | 2,382 | +65% | 0 | 0 | — |
case-19 | pass→pass | 10,713 | 12,579 | +17% | 1 | 1 | 0% | 1,763 | 2,654 | +51% | 0 | 0 | — |
case-20 | pass→pass | 12,553 | 7,208 | -43% | 1 | 1 | 0% | 2,086 | 1,889 | -9% | 0 | 0 | — |
case-21 | fail→fail | 10,630 | 12,890 | +21% | 1 | 1 | 0% | 1,820 | 2,771 | +52% | 0 | 0 | — |
case-22 | fail→pass | 11,013 | 13,643 | +24% | 1 | 1 | 0% | 1,892 | 2,898 | +53% | 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 +14 percentage points is the difference between those two pass rates over the 22 comparable cases. 3 cases got worse with the skill loaded, and they are 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.