Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Release designer-skill-mcp to npm with a git tag and GitHub release. Use when the user asks to ship, publish, cut a release, bump version, push tags, or run npm publish for designer-skill-mcp.
.claude/skills/hashgraph-online-release/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -36% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -50% | 0% |
Use only when the user asks to release.
If topology=multi-module (triage announcement or coordinator spec/plan declaration), read ../references/multi-module.md. Release only the exact coordinator revision set that passed integration. Determine version and dependency order per repository, and keep the existing approval gates for each repository's commit, tag, push, or publish. A completed change set does not itself authorize a release.
Unreleased to version - date; create new Unreleased.<gate> Steps 5-7 touch shared or irreversible state. Get explicit user approval before each step — never chain them on one confirmation. </gate>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 12,650 | 10,994 | -13% | 1 | 1 | 0% | 1,209 | 633 | -48% | 0 | 0 | — |
case-02 | fail→fail | 2,364 | 3,243 | +37% | 1 | 1 | 0% | 259 | 566 | +119% | 0 | 0 | — |
case-03 | fail→fail | 15,998 | 7,611 | -52% | 1 | 1 | 0% | 1,530 | 632 | -59% | 0 | 0 | — |
case-04 | fail→pass | 8,240 | 8,760 | +6% | 1 | 1 | 0% | 1,205 | 777 | -36% | 0 | 0 | — |
case-05 | pass→pass | 18,225 | 12,128 | -33% | 1 | 1 | 0% | 2,114 | 1,237 | -41% | 0 | 0 | — |
case-06 | pass→pass | 14,122 | 8,435 | -40% | 1 | 1 | 0% | 1,257 | 828 | -34% | 0 | 0 | — |
case-07 | pass→pass | 11,843 | 8,320 | -30% | 1 | 1 | 0% | 1,103 | 781 | -29% | 0 | 0 | — |
case-08 | fail→pass | 19,776 | 9,826 | -50% | 1 | 1 | 0% | 1,798 | 826 | -54% | 0 | 0 | — |
case-09 | fail→pass | 8,222 | 7,281 | -11% | 1 | 1 | 0% | 1,259 | 408 | -68% | 0 | 0 | — |
case-10 | pass→pass | 24,217 | 8,185 | -66% | 1 | 1 | 0% | 2,378 | 764 | -68% | 0 | 0 | — |
case-11 | fail→pass | 16,346 | 5,417 | -67% | 1 | 1 | 0% | 1,808 | 952 | -47% | 0 | 0 | — |
case-12 | fail→pass | 11,787 | 3,398 | -71% | 1 | 1 | 0% | 1,676 | 842 | -50% | 0 | 0 | — |
case-13 | fail→pass | 11,679 | 9,534 | -18% | 1 | 1 | 0% | 1,076 | 782 | -27% | 0 | 0 | — |
case-14 | pass→pass | 13,116 | 8,429 | -36% | 1 | 1 | 0% | 1,055 | 764 | -28% | 0 | 0 | — |
case-15 | pass→pass | 14,123 | 8,876 | -37% | 1 | 1 | 0% | 1,283 | 924 | -28% | 0 | 0 | — |
case-16 | pass→pass | 20,114 | 7,534 | -63% | 1 | 1 | 0% | 3,239 | 1,174 | -64% | 0 | 0 | — |
case-17 | fail→pass | 17,397 | 9,408 | -46% | 1 | 1 | 0% | 1,905 | 1,628 | -15% | 0 | 0 | — |
case-18 | fail→pass | 17,031 | 11,867 | -30% | 1 | 1 | 0% | 1,991 | 1,131 | -43% | 0 | 0 | — |
case-19 | fail→pass | 13,621 | 3,701 | -73% | 1 | 1 | 0% | 1,914 | 730 | -62% | 0 | 0 | — |
case-20 | pass→pass | 9,383 | 7,340 | -22% | 1 | 1 | 0% | 581 | 573 | -1% | 0 | 0 | — |
case-21 | pass→pass | 15,307 | 6,856 | -55% | 1 | 1 | 0% | 1,565 | 488 | -69% | 0 | 0 | — |
case-22 | pass→pass | 11,324 | 9,264 | -18% | 1 | 1 | 0% | 1,663 | 819 | -51% | 0 | 0 | — |
case-23 | pass→pass | 8,539 | 6,106 | -28% | 1 | 1 | 0% | 1,495 | 1,078 | -28% | 0 | 0 | — |
case-24 | pass→pass | 4,649 | 4,280 | -8% | 1 | 1 | 0% | 820 | 1,036 | +26% | 0 | 0 | — |
case-25 | pass→pass | 18,355 | 25,982 | +42% | 1 | 1 | 0% | 2,791 | 3,277 | +17% | 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. 25 cases were attempted, and 24 counted toward the lift figure. The other 1 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 +36 percentage points is the difference between those two pass rates over the 24 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.