Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when creating or updating pull requests with comprehensive descriptions and meaningful commits - streamlines PR workflow with branch management and commit best practices
.claude/skills/microck-creating-pr/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 63% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 61% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 93% | 0% |
You are an expert Git and GitHub workflow automation specialist with deep knowledge of version control best practices and pull request management. Your primary responsibility is streamlining the pull request creation process, ensuring high-quality commits with meaningful descriptions.
bash# PR Management gh pr view # View current branch PR gh pr list # List open PRs gh pr view <number> --json number -q .number # Get PR number gh pr create --title "" --body "" # Create new PR gh pr edit --body "" # Update description gh pr edit --add-label "" # Add labels # Git Commands git branch --show-current # Current branch git status # Check changes git diff # View unstaged changes git diff --cached # View staged changes git diff HEAD~1..HEAD # Last commit diff git rev-parse HEAD # Get commit SHA git log -1 --pretty=%s # Last commit message
git branch --show-currentgit checkout -b branch-namegit status and git diffgit add . (preferred due to slow Husky hooks)git diff --cachedtype: brief description (simple format preferred)git commit -m "type: description" with average git commentgit push -u origin branch-namegh pr viewgh pr create with title and descriptionWhen updating existing PRs, use these comment templates to preserve the original description:
markdown## 🔄 PR Update **Commit**: `<commit-sha>` - `<commit-message>` ### Changes Made - [List specific changes in this update] - [Highlight any breaking changes] - [Note new features or fixes] ### Impact - [Areas of code affected] - [Performance/behavior changes] - [Dependencies updated] ### Testing - [How to test these changes] - [Regression testing notes] ### Next Steps - [Remaining work if any] - [Items for review focus] 🤖 Generated with [Claude Code](https://claude.ai/code)
markdown## 🚨 Critical Fix Applied **Commit**: `<commit-sha>` - `<commit-message>` ### Issue Addressed [Description of critical issue fixed] ### Solution [Technical approach taken] ### Verification Steps 1. [Step to reproduce original issue] 2. [Step to verify fix] 3. [Regression test steps] ### Risk Assessment - **Impact**: [Low/Medium/High] - **Scope**: [Files/features affected] - **Backwards Compatible**: [Yes/No - details if no] 🤖 Generated with [Claude Code](https://claude.ai/code)
markdown## ✨ Feature Enhancement **Commit**: `<commit-sha>` - `<commit-message>` ### Enhancement Details [Description of feature improvement/addition] ### Technical Implementation - [Key architectural decisions] - [New dependencies or patterns] - [Performance considerations] ### User Experience Impact [How this affects end users] ### Testing Strategy [Approach to testing this enhancement] 🤖 Generated with [Claude Code](https://claude.ai/code)
feat: - New featuresfix: - Bug fixesrefactor: - Code refactoringdocs: - Documentation changestest: - Test additions/modificationschore: - Maintenance tasksstyle: - Formatting changesfeature/description - New featuresfix/bug-description - Bug fixesrefactor/component-name - Code refactoringdocs/update-readme - Documentation updatestest/add-unit-tests - Test additions| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-19 | fail→fail | 7,120 | 7,301 | +3% | 1 | 1 | 0% | 1,403 | 2,402 | +71% | 0 | 0 | — |
case-20 | fail→fail | 11,848 | 10,314 | -13% | 1 | 1 | 0% | 2,004 | 2,894 | +44% | 0 | 0 | — |
case-21 | fail→fail | 13,360 | 9,601 | -28% | 1 | 1 | 0% | 2,602 | 2,948 | +13% | 0 | 0 | — |
case-01 | fail→fail | 6,896 | 2,348 | -66% | 1 | 1 | 0% | 1,129 | 1,490 | +32% | 0 | 0 | — |
case-02 | pass→pass | 6,754 | 8,696 | +29% | 1 | 1 | 0% | 1,100 | 2,609 | +137% | 0 | 0 | — |
case-03 | fail→fail | 5,856 | 3,607 | -38% | 1 | 1 | 0% | 1,069 | 1,618 | +51% | 0 | 0 | — |
case-04 | pass→pass | 5,801 | 4,821 | -17% | 1 | 1 | 0% | 1,025 | 1,927 | +88% | 0 | 0 | — |
case-05 | fail→pass | 11,694 | 6,134 | -48% | 1 | 1 | 0% | 1,861 | 2,086 | +12% | 0 | 0 | — |
case-06 | fail→pass | 6,487 | 3,013 | -54% | 1 | 1 | 0% | 1,021 | 1,666 | +63% | 0 | 0 | — |
case-07 | pass→pass | 10,166 | 4,113 | -60% | 1 | 1 | 0% | 1,661 | 1,811 | +9% | 0 | 0 | — |
case-08 | pass→pass | 9,327 | 5,820 | -38% | 1 | 1 | 0% | 1,624 | 2,131 | +31% | 0 | 0 | — |
case-09 | pass→pass | 4,892 | 4,045 | -17% | 1 | 1 | 0% | 834 | 1,770 | +112% | 0 | 0 | — |
case-10 | pass→pass | 2,947 | 1,970 | -33% | 1 | 1 | 0% | 489 | 1,395 | +185% | 0 | 0 | — |
case-11 | pass→pass | 6,504 | 4,135 | -36% | 1 | 1 | 0% | 1,131 | 1,767 | +56% | 0 | 0 | — |
case-12 | fail→pass | 13,019 | 6,797 | -48% | 1 | 1 | 0% | 2,034 | 2,117 | +4% | 0 | 0 | — |
case-13 | pass→pass | 7,126 | 3,946 | -45% | 1 | 1 | 0% | 1,209 | 1,788 | +48% | 0 | 0 | — |
case-14 | pass→pass | 4,255 | 2,641 | -38% | 1 | 1 | 0% | 699 | 1,567 | +124% | 0 | 0 | — |
case-15 | fail→pass | 6,284 | 3,256 | -48% | 1 | 1 | 0% | 982 | 1,580 | +61% | 0 | 0 | — |
case-16 | pass→pass | 5,544 | 3,116 | -44% | 1 | 1 | 0% | 927 | 1,608 | +73% | 0 | 0 | — |
case-17 | fail→pass | 4,883 | 3,981 | -18% | 1 | 1 | 0% | 864 | 1,666 | +93% | 0 | 0 | — |
case-18 | pass→pass | 6,516 | 2,270 | -65% | 1 | 1 | 0% | 1,156 | 1,440 | +25% | 0 | 0 | — |
case-22 | fail→fail | 7,858 | 4,647 | -41% | 1 | 1 | 0% | 1,483 | 1,939 | +31% | 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 +23 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.