Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →PR creation, CI/CD validation, merge coordination, and release patterns. Use when creating pull requests, running pre-PR validation, checking CI status, coordinating merges, or managing releases. Do NOT use for routine development commits -- see safe-workflow skill instead.
.claude/skills/bybren-llc-release-patterns/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -13% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -56% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -42% | 0% |
Ensure consistent PR creation, CI/CD validation, and release coordination following rebase-first workflow.
Before creating any PR:
{{TICKET_PREFIX}}-{number}-{description}type(scope): description [{{TICKET_PREFIX}}-XXX]git fetch origin && git rebase origin/{{MAIN_BRANCH}}{{CI_VALIDATE_COMMAND}}bashgh pr create \ --title "feat(scope): description [{{TICKET_PREFIX}}-XXX]" \ --body "## Summary Implements feature as specified in {{TICKET_PREFIX}}-XXX. ## Changes - Change 1 - Change 2 ## Testing - CI passes - Manual testing completed"
ONLY use rebase merge:
bash# CORRECT gh pr merge --rebase --delete-branch # NEVER gh pr merge --squash # Loses history gh pr merge --merge # Creates merge commits
.github/pull_request_template.mdCONTRIBUTING.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 | 10,060 | 6,475 | -36% | 1 | 1 | 0% | 1,745 | 1,525 | -13% | 0 | 0 | — |
case-02 | fail→pass | 9,152 | 6,080 | -34% | 1 | 1 | 0% | 1,521 | 1,415 | -7% | 0 | 0 | — |
case-03 | fail→pass | 9,972 | 2,533 | -75% | 1 | 1 | 0% | 1,633 | 712 | -56% | 0 | 0 | — |
case-04 | fail→pass | 6,935 | 2,675 | -61% | 1 | 1 | 0% | 1,089 | 767 | -30% | 0 | 0 | — |
case-05 | fail→pass | 9,944 | 3,485 | -65% | 1 | 1 | 0% | 1,655 | 954 | -42% | 0 | 0 | — |
case-06 | pass→pass | 10,196 | 4,612 | -55% | 1 | 1 | 0% | 1,779 | 1,080 | -39% | 0 | 0 | — |
case-07 | fail→pass | 8,388 | 3,919 | -53% | 1 | 1 | 0% | 1,409 | 1,016 | -28% | 0 | 0 | — |
case-08 | fail→fail | 6,714 | 1,950 | -71% | 1 | 1 | 0% | 1,074 | 631 | -41% | 0 | 0 | — |
case-09 | fail→pass | 6,408 | 2,479 | -61% | 1 | 1 | 0% | 954 | 741 | -22% | 0 | 0 | — |
case-10 | fail→pass | 11,900 | 2,449 | -79% | 1 | 1 | 0% | 1,758 | 716 | -59% | 0 | 0 | — |
case-11 | pass→pass | 5,071 | 2,069 | -59% | 1 | 1 | 0% | 804 | 622 | -23% | 0 | 0 | — |
case-12 | pass→pass | 5,591 | 1,508 | -73% | 1 | 1 | 0% | 960 | 512 | -47% | 0 | 0 | — |
case-13 | pass→pass | 3,846 | 1,503 | -61% | 1 | 1 | 0% | 569 | 495 | -13% | 0 | 0 | — |
case-14 | fail→pass | 10,397 | 4,002 | -62% | 1 | 1 | 0% | 1,440 | 966 | -33% | 0 | 0 | — |
case-15 | pass→pass | 6,644 | 3,746 | -44% | 1 | 1 | 0% | 1,112 | 931 | -16% | 0 | 0 | — |
case-16 | pass→pass | 14,588 | 5,345 | -63% | 1 | 1 | 0% | 2,097 | 1,047 | -50% | 0 | 0 | — |
case-17 | fail→pass | 9,474 | 2,130 | -78% | 1 | 1 | 0% | 1,319 | 641 | -51% | 0 | 0 | — |
case-18 | pass→pass | 9,243 | 2,515 | -73% | 1 | 1 | 0% | 1,626 | 724 | -55% | 0 | 0 | — |
case-19 | pass→pass | 2,763 | 2,103 | -24% | 1 | 1 | 0% | 396 | 626 | +58% | 0 | 0 | — |
case-20 | pass→pass | 10,419 | 7,112 | -32% | 1 | 1 | 0% | 1,744 | 1,609 | -8% | 0 | 0 | — |
case-21 | pass→pass | 6,873 | 5,084 | -26% | 1 | 1 | 0% | 1,171 | 1,147 | -2% | 0 | 0 | — |
case-22 | pass→pass | 15,364 | 13,099 | -15% | 1 | 1 | 0% | 2,731 | 2,683 | -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. 22 cases were attempted. The headline lift of +45 percentage points is the difference between those two pass rates over the 22 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.