Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Release preparation — changelog generation, version bump, tag creation. Generic version (not project-specific). For EvoNexus releases, use custom-release instead.
.claude/skills/evolution-foundation-dev-release/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 542% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -19% | 0% |
Derived from oh-my-claudecode (MIT, Yeachan Heo). Adapted for the EvoNexus Engineering Layer.
Generic release preparation: changelog generation from git log, version bump, tag creation. For EvoNexus-specific releases, use custom-release (the existing skill that handles git-flow develop→main).
custom-release insteadmain/master directly)dev-verify)package.json, Cargo.toml, go.mod, pyproject.toml)git log {last-tag}..HEAD --onelineCHANGELOG.md with new version sectiongit commit -m "chore(release): vX.Y.Z"git tag vX.Y.Zgit push origin vX.Y.Z (after user confirmation)@oath-verifier confirms the release commit and tag are correctSave release notes to workspace/development/research/[C]release-{version}-{date}.md.
@flow-git (commits and tags)@oath-verifier (verification)@quill-writer (changelog formatting)dev-verify (pre-flight)EvoNexus has its own release skill (custom-release) that handles the git-flow develop→main workflow with EvoNexus-specific gates (CHANGELOG entry, version sync across files, GitHub release creation). Use custom-release for EvoNexus releases. Use dev-release only for projects in workspace/projects/ that have their own release lifecycle.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | fail→pass | 5,116 | 4,508 | -12% | 1 | 1 | 0% | 203 | 1,303 | +542% | 0 | 0 | — |
case-01 | fail→fail | 4,520 | 3,061 | -32% | 1 | 1 | 0% | 272 | 989 | +264% | 0 | 0 | — |
case-02 | fail→fail | 4,709 | 3,126 | -34% | 1 | 1 | 0% | 322 | 1,039 | +223% | 0 | 0 | — |
case-03 | fail→fail | 5,733 | 6,352 | +11% | 1 | 1 | 0% | 270 | 949 | +251% | 0 | 0 | — |
case-04 | pass→pass | 11,077 | 3,102 | -72% | 1 | 1 | 0% | 1,691 | 1,182 | -30% | 0 | 0 | — |
case-05 | pass→pass | 9,915 | 3,417 | -66% | 1 | 1 | 0% | 1,702 | 1,107 | -35% | 0 | 0 | — |
case-06 | pass→pass | 8,616 | 4,709 | -45% | 1 | 1 | 0% | 1,417 | 1,386 | -2% | 0 | 0 | — |
case-07 | fail→pass | 10,301 | 5,616 | -45% | 1 | 1 | 0% | 1,877 | 1,707 | -9% | 0 | 0 | — |
case-08 | pass→pass | 6,753 | 5,048 | -25% | 1 | 1 | 0% | 1,159 | 1,426 | +23% | 0 | 0 | — |
case-10 | fail→pass | 5,404 | 3,316 | -39% | 1 | 1 | 0% | 956 | 1,210 | +27% | 0 | 0 | — |
case-11 | pass→pass | 13,849 | 7,330 | -47% | 1 | 1 | 0% | 2,432 | 1,848 | -24% | 0 | 0 | — |
case-12 | fail→pass | 4,968 | 1,711 | -66% | 1 | 1 | 0% | 924 | 863 | -7% | 0 | 0 | — |
case-13 | pass→pass | 3,217 | 3,735 | +16% | 1 | 1 | 0% | 547 | 1,203 | +120% | 0 | 0 | — |
case-14 | pass→pass | 3,791 | 2,946 | -22% | 1 | 1 | 0% | 584 | 1,032 | +77% | 0 | 0 | — |
case-15 | pass→pass | 6,906 | 3,365 | -51% | 1 | 1 | 0% | 1,114 | 1,129 | +1% | 0 | 0 | — |
case-16 | pass→fail | 5,335 | 2,465 | -54% | 1 | 1 | 0% | 886 | 983 | +11% | 0 | 0 | — |
case-17 | fail→fail | 9,326 | 2,272 | -76% | 1 | 1 | 0% | 1,632 | 962 | -41% | 0 | 0 | — |
case-18 | fail→pass | 6,105 | 1,279 | -79% | 1 | 1 | 0% | 922 | 745 | -19% | 0 | 0 | — |
case-19 | fail→pass | 13,050 | 1,122 | -91% | 1 | 1 | 0% | 2,271 | 762 | -66% | 0 | 0 | — |
case-20 | fail→fail | 7,699 | 2,080 | -73% | 1 | 1 | 0% | 1,189 | 851 | -28% | 0 | 0 | — |
case-21 | fail→fail | 10,178 | 1,493 | -85% | 1 | 1 | 0% | 1,682 | 794 | -53% | 0 | 0 | — |
case-22 | fail→fail | 9,067 | 1,542 | -83% | 1 | 1 | 0% | 1,522 | 794 | -48% | 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 18 counted toward the lift figure. The other 4 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 +23 percentage points is the difference between those two pass rates over the 18 comparable cases. 1 case got worse with the skill loaded, and it is 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.