Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Triage support tickets by pulling helpdesk context, searching issue trackers and knowledge bases in parallel, classifying the issue, and giving reps a recommended next action.
.claude/skills/zapier-support-ticket-triage/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 90% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 41% | 0% |
Use this skill when a support rep needs fast, source-backed context before replying to a customer ticket.
This skill is generalized from a real implementation by Corey Smith at ClickUp. Corey built an MCP-powered support triage agent that pulls Zendesk context, searches defect tracking and feature backlog systems in parallel, classifies the ticket, and gives reps structured context before they type a reply.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 14,110 | 13,256 | -6% | 1 | 1 | 0% | 2,223 | 2,673 | +20% | 0 | 0 | — |
case-02 | fail→pass | 14,634 | 12,538 | -14% | 1 | 1 | 0% | 2,289 | 2,568 | +12% | 0 | 0 | — |
case-03 | fail→pass | 12,114 | 11,166 | -8% | 1 | 1 | 0% | 1,788 | 2,234 | +25% | 0 | 0 | — |
case-04 | fail→fail | 10,681 | 9,221 | -14% | 1 | 1 | 0% | 1,663 | 1,919 | +15% | 0 | 0 | — |
case-05 | pass→pass | 11,049 | 9,268 | -16% | 1 | 1 | 0% | 1,653 | 1,944 | +18% | 0 | 0 | — |
case-06 | fail→pass | 6,234 | 8,420 | +35% | 1 | 1 | 0% | 948 | 1,804 | +90% | 0 | 0 | — |
case-07 | pass→pass | 6,410 | 6,205 | -3% | 1 | 1 | 0% | 968 | 1,510 | +56% | 0 | 0 | — |
case-08 | fail→pass | 6,910 | 6,299 | -9% | 1 | 1 | 0% | 1,110 | 1,568 | +41% | 0 | 0 | — |
case-09 | fail→pass | 13,943 | 8,503 | -39% | 1 | 1 | 0% | 2,253 | 1,870 | -17% | 0 | 0 | — |
case-10 | fail→pass | 18,260 | 10,498 | -43% | 1 | 1 | 0% | 2,404 | 2,245 | -7% | 0 | 0 | — |
case-11 | pass→pass | 8,542 | 9,086 | +6% | 1 | 1 | 0% | 1,300 | 1,927 | +48% | 0 | 0 | — |
case-12 | fail→pass | 7,927 | 3,570 | -55% | 1 | 1 | 0% | 1,285 | 1,073 | -16% | 0 | 0 | — |
case-13 | fail→pass | 9,635 | 3,547 | -63% | 1 | 1 | 0% | 1,437 | 1,098 | -24% | 0 | 0 | — |
case-14 | pass→pass | 7,813 | 9,609 | +23% | 1 | 1 | 0% | 1,177 | 1,953 | +66% | 0 | 0 | — |
case-15 | pass→pass | 8,756 | 9,505 | +9% | 1 | 1 | 0% | 1,264 | 1,912 | +51% | 0 | 0 | — |
case-16 | pass→pass | 7,783 | 8,531 | +10% | 1 | 1 | 0% | 1,165 | 1,863 | +60% | 0 | 0 | — |
case-17 | fail→pass | 13,604 | 3,896 | -71% | 1 | 1 | 0% | 1,918 | 1,019 | -47% | 0 | 0 | — |
case-18 | fail→pass | 11,468 | 4,266 | -63% | 1 | 1 | 0% | 1,726 | 1,225 | -29% | 0 | 0 | — |
case-19 | fail→pass | 4,993 | 9,051 | +81% | 1 | 1 | 0% | 731 | 1,923 | +163% | 0 | 0 | — |
case-20 | fail→fail | 8,249 | 9,911 | +20% | 1 | 1 | 0% | 1,487 | 2,274 | +53% | 0 | 0 | — |
case-21 | pass→pass | 6,311 | 8,275 | +31% | 1 | 1 | 0% | 933 | 1,688 | +81% | 0 | 0 | — |
case-22 | pass→pass | 9,883 | 10,707 | +8% | 1 | 1 | 0% | 1,594 | 2,059 | +29% | 0 | 0 | — |
case-23 | pass→pass | 6,764 | 3,688 | -45% | 1 | 1 | 0% | 1,027 | 1,052 | +2% | 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 +52 percentage points is the difference between those two pass rates over the 23 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.