Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Aggregate and analyze customer reviews from G2, Capterra, Trustpilot, App Store, and other platforms. Performs sentiment analysis, identifies pain points, extracts feature feedback, generates marketing claims, and compares competitor reviews. Use when users need review analysis, competitive intelligence, or customer feedback insights.
.claude/skills/onewave-ai-customer-review-aggregator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -51% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 14% | 0% |
Pull reviews from multiple platforms and extract actionable insights with sentiment analysis, pain-point detection, marketing-claim extraction, and competitor comparison.
references/sources.md - supported platforms and sentiment dimensionsreferences/intake-prompts.md - scope and data-collection prompts, example use casesreferences/output-templates.md - report templates for every analysis typereferences/intake-prompts.md to capture product, platforms, competitors, analysis focus, and time period.references/intake-prompts.md and collect the reviews.references/sources.md.For all output formats and tables, see references/output-templates.md.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 19,256 | 29,390 | +53% | 1 | 1 | 0% | 2,965 | 1,452 | -51% | 0 | 0 | — |
case-02 | fail→pass | 14,704 | 6,937 | -53% | 1 | 1 | 0% | 1,998 | 1,429 | -28% | 0 | 0 | — |
case-03 | fail→pass | 6,476 | 8,308 | +28% | 1 | 1 | 0% | 1,052 | 1,523 | +45% | 0 | 0 | — |
case-04 | fail→fail | 10,400 | 15,134 | +46% | 1 | 1 | 0% | 982 | 2,209 | +125% | 0 | 0 | — |
case-05 | fail→fail | 14,309 | 13,476 | -6% | 1 | 1 | 0% | 2,028 | 2,084 | +3% | 0 | 0 | — |
case-06 | fail→fail | 16,462 | 14,186 | -14% | 1 | 1 | 0% | 2,410 | 2,635 | +9% | 0 | 0 | — |
case-07 | fail→fail | 21,629 | 25,315 | +17% | 1 | 1 | 0% | 4,129 | 5,192 | +26% | 0 | 0 | — |
case-08 | pass→pass | 11,309 | 5,997 | -47% | 1 | 1 | 0% | 1,759 | 1,265 | -28% | 0 | 0 | — |
case-09 | pass→fail | 9,129 | 5,757 | -37% | 1 | 1 | 0% | 1,311 | 1,175 | -10% | 0 | 0 | — |
case-10 | pass→pass | 14,398 | 14,016 | -3% | 1 | 1 | 0% | 2,238 | 2,534 | +13% | 0 | 0 | — |
case-11 | fail→pass | 16,790 | 11,003 | -34% | 1 | 1 | 0% | 2,187 | 1,936 | -11% | 0 | 0 | — |
case-12 | fail→pass | 12,528 | 12,468 | -0% | 1 | 1 | 0% | 1,959 | 2,234 | +14% | 0 | 0 | — |
case-13 | pass→fail | 15,812 | 16,598 | +5% | 1 | 1 | 0% | 2,371 | 2,499 | +5% | 0 | 0 | — |
case-14 | fail→fail | 13,641 | 11,676 | -14% | 1 | 1 | 0% | 2,032 | 1,862 | -8% | 0 | 0 | — |
case-15 | pass→fail | 5,167 | 9,148 | +77% | 1 | 1 | 0% | 685 | 1,696 | +148% | 0 | 0 | — |
case-16 | fail→pass | 17,877 | 16,470 | -8% | 1 | 1 | 0% | 2,629 | 2,568 | -2% | 0 | 0 | — |
case-17 | fail→fail | 15,402 | 10,478 | -32% | 1 | 1 | 0% | 2,276 | 1,886 | -17% | 0 | 0 | — |
case-18 | fail→pass | 15,642 | 14,478 | -7% | 1 | 1 | 0% | 2,229 | 2,375 | +7% | 0 | 0 | — |
case-19 | pass→pass | 16,501 | 9,882 | -40% | 1 | 1 | 0% | 2,484 | 1,894 | -24% | 0 | 0 | — |
case-20 | fail→pass | 15,503 | 11,656 | -25% | 1 | 1 | 0% | 2,441 | 2,233 | -9% | 0 | 0 | — |
case-21 | pass→pass | 13,297 | 9,776 | -26% | 1 | 1 | 0% | 1,952 | 1,680 | -14% | 0 | 0 | — |
case-22 | pass→pass | 14,506 | 14,523 | +0% | 1 | 1 | 0% | 1,932 | 2,355 | +22% | 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 +23 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.