Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when cutting a GodotPrompter release or bumping its version — the version-bump sequence, tag-triggered workflow, and the marketplace manifests that must follow.
.claude/skills/jame581-releasing-godot-prompter/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -4% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -22% | 0% |
Current version: check package.json and .claude-plugin/plugin.json (must match). Full command sequence lives in CONTRIBUTING.md.
node scripts/bump-version.mjs <version> — bumps package.json,.claude-plugin/plugin.json, .claude-plugin/marketplace.json, plus sibling marketplaces if present; also syncs the live skill count into the "N domain-specific skills" text of each manifest description.
CHANGELOG.md with the new section.v<version>), push with tags — .github/workflows/release.ymlthen validates, creates the GitHub release, and opens marketplace PRs (when MARKETPLACE_TOKEN is configured).
skillsmith/.claude-plugin/marketplace.json (primary) and the legacy godot-prompter-marketplace/.claude-plugin/marketplace.json.
plugin.json (Antigravity) must match package.json too — step 1handles it, but verify before tagging.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 9,322 | 6,638 | -29% | 1 | 1 | 0% | 1,708 | 1,632 | -4% | 0 | 0 | — |
case-02 | fail→pass | 21,225 | 5,683 | -73% | 1 | 1 | 0% | 1,026 | 1,299 | +27% | 0 | 0 | — |
case-03 | fail→pass | 15,457 | 8,888 | -42% | 1 | 1 | 0% | 2,473 | 2,037 | -18% | 0 | 0 | — |
case-04 | pass→pass | 17,081 | 16,509 | -3% | 1 | 1 | 0% | 3,174 | 3,439 | +8% | 0 | 0 | — |
case-05 | pass→pass | 11,798 | 8,827 | -25% | 1 | 1 | 0% | 2,130 | 1,857 | -13% | 0 | 0 | — |
case-06 | fail→pass | 16,945 | 15,452 | -9% | 1 | 1 | 0% | 3,157 | 3,262 | +3% | 0 | 0 | — |
case-07 | fail→pass | 9,061 | 4,703 | -48% | 1 | 1 | 0% | 1,564 | 1,215 | -22% | 0 | 0 | — |
case-08 | fail→pass | 8,044 | 3,062 | -62% | 1 | 1 | 0% | 1,263 | 834 | -34% | 0 | 0 | — |
case-09 | fail→pass | 9,516 | 4,840 | -49% | 1 | 1 | 0% | 1,692 | 1,238 | -27% | 0 | 0 | — |
case-10 | pass→pass | 9,235 | 4,180 | -55% | 1 | 1 | 0% | 1,504 | 1,091 | -27% | 0 | 0 | — |
case-11 | pass→pass | 8,504 | 2,287 | -73% | 1 | 1 | 0% | 1,551 | 697 | -55% | 0 | 0 | — |
case-12 | pass→pass | 6,195 | 1,720 | -72% | 1 | 1 | 0% | 988 | 529 | -46% | 0 | 0 | — |
case-13 | fail→pass | 10,953 | 1,537 | -86% | 1 | 1 | 0% | 1,864 | 518 | -72% | 0 | 0 | — |
case-14 | fail→pass | 5,191 | 1,999 | -61% | 1 | 1 | 0% | 821 | 659 | -20% | 0 | 0 | — |
case-15 | fail→pass | 11,103 | 3,159 | -72% | 1 | 1 | 0% | 1,664 | 833 | -50% | 0 | 0 | — |
case-16 | fail→pass | 8,435 | 2,153 | -74% | 1 | 1 | 0% | 1,279 | 632 | -51% | 0 | 0 | — |
case-17 | fail→pass | 17,263 | 1,817 | -89% | 1 | 1 | 0% | 2,845 | 563 | -80% | 0 | 0 | — |
case-18 | fail→pass | 13,548 | 1,721 | -87% | 1 | 1 | 0% | 2,103 | 543 | -74% | 0 | 0 | — |
case-19 | fail→pass | 7,431 | 1,335 | -82% | 1 | 1 | 0% | 1,126 | 479 | -57% | 0 | 0 | — |
case-20 | fail→pass | 10,348 | 3,920 | -62% | 1 | 1 | 0% | 1,784 | 1,058 | -41% | 0 | 0 | — |
case-21 | fail→fail | 9,902 | 5,490 | -45% | 1 | 1 | 0% | 1,885 | 1,387 | -26% | 0 | 0 | — |
case-22 | fail→pass | 11,703 | 1,783 | -85% | 1 | 1 | 0% | 1,975 | 625 | -68% | 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, and 21 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 +73 percentage points is the difference between those two pass rates over the 21 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.