Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Guide for creating comprehensive PR descriptions with proper structure, diagrams, and documentation for code reviews.
.claude/skills/aiskillstore-pr-description/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 240% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 99% | 0% |
This skill provides a structured approach to creating comprehensive pull request descriptions that help reviewers understand changes quickly.
Use this skill when:
If saving the PR description to a file:
pr_description_<descriptive_title>_<timestamp>.mddate +"%Y%m%d_%H%M%S"pr_description_user_auth_refactor_20250612_152832.mdmarkdown# PR Title: [Descriptive Title] ## 📋 Summary Brief overview of what this PR accomplishes in 2-3 sentences. ## 📊 Data Flow Diagram [Mermaid diagram showing the flow/architecture - use `mermaid` skill for styling] ## 🗂️ Entity Relationship Diagram [Include if creating/modifying data relationships - Mermaid ERD] ## 🎯 Goals - Primary objective 1 - Primary objective 2 - Any secondary objectives ## 🔍 Changes in this PR - Description of key changes - Any config file changes - Schema modifications - Breaking changes (if any) ## 🧪 Testing ### Commands Run
npm test npm run lint
### Results
- ✅ All tests passed
- ✅ Linting passed
- ✅ Build successful
- Include any relevant screenshots or output
### Validation
- Key scenarios tested
- Edge cases covered
- Performance considerations
## 📋 Notes for Reviewers
- Pay attention to: [specific areas]
- Assumptions made: [list any assumptions]
- Dependencies: [any upstream/downstream impacts]
## 🔜 Future Work
- Any follow-up tasks not included in this PR
- Potential optimisations
- Additional testing neededUse the mermaid skill for diagram syntax and styling that works in both light and dark mode.
Show the complete flow of data or logic:
Show object/data relationships:
"This PR refactors the user authentication flow to use JWT tokens instead of session cookies. The change improves scalability for our microservices architecture and reduces server-side session storage requirements."
bash# Commands run npm run test:unit npm run test:integration npm run lint
Results:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-07 | fail→pass | 11,600 | 20,327 | +75% | 1 | 1 | 0% | 914 | 3,107 | +240% | 0 | 0 | — |
case-13 | fail→pass | 21,211 | 32,009 | +51% | 1 | 1 | 0% | 2,889 | 3,709 | +28% | 0 | 0 | — |
case-01 | pass→pass | 9,990 | 8,871 | -11% | 1 | 1 | 0% | 844 | 1,563 | +85% | 0 | 0 | — |
case-02 | pass→pass | 13,517 | 16,464 | +22% | 1 | 1 | 0% | 1,525 | 2,469 | +62% | 0 | 0 | — |
case-03 | pass→pass | 14,560 | 10,133 | -30% | 1 | 1 | 0% | 1,533 | 2,555 | +67% | 0 | 0 | — |
case-04 | fail→fail | 22,728 | 22,097 | -3% | 1 | 1 | 0% | 3,557 | 2,782 | -22% | 0 | 0 | — |
case-05 | fail→pass | 22,517 | 15,703 | -30% | 1 | 1 | 0% | 2,656 | 3,507 | +32% | 0 | 0 | — |
case-06 | fail→pass | 19,044 | 23,879 | +25% | 1 | 1 | 0% | 3,076 | 3,230 | +5% | 0 | 0 | — |
case-08 | pass→fail | 11,804 | 35,130 | +198% | 1 | 1 | 0% | 976 | 1,161 | +19% | 0 | 0 | — |
case-09 | fail→fail | 20,362 | 27,814 | +37% | 1 | 1 | 0% | 2,609 | 3,918 | +50% | 0 | 0 | — |
case-10 | fail→fail | 15,238 | 46,938 | +208% | 1 | 1 | 0% | 1,775 | 1,523 | -14% | 0 | 0 | — |
case-11 | fail→fail | 32,815 | 35,916 | +9% | 1 | 1 | 0% | 2,835 | 4,580 | +62% | 0 | 0 | — |
case-12 | pass→pass | 23,988 | 13,449 | -44% | 1 | 1 | 0% | 1,392 | 3,215 | +131% | 0 | 0 | — |
case-14 | fail→pass | 24,100 | 13,728 | -43% | 1 | 1 | 0% | 1,245 | 2,480 | +99% | 0 | 0 | — |
case-15 | fail→pass | 26,360 | 24,574 | -7% | 1 | 1 | 0% | 1,781 | 3,127 | +76% | 0 | 0 | — |
case-16 | fail→pass | 31,060 | 32,484 | +5% | 1 | 1 | 0% | 1,487 | 2,527 | +70% | 0 | 0 | — |
case-17 | pass→pass | 30,754 | 33,482 | +9% | 1 | 1 | 0% | 1,597 | 3,930 | +146% | 0 | 0 | — |
case-18 | pass→pass | 41,432 | 34,851 | -16% | 1 | 1 | 0% | 3,130 | 4,740 | +51% | 0 | 0 | — |
case-19 | fail→pass | 34,140 | 39,805 | +17% | 1 | 1 | 0% | 1,945 | 4,363 | +124% | 0 | 0 | — |
case-20 | fail→pass | 29,844 | 33,911 | +14% | 1 | 1 | 0% | 1,760 | 3,158 | +79% | 0 | 0 | — |
case-21 | pass→pass | 30,141 | 21,108 | -30% | 1 | 1 | 0% | 1,599 | 3,921 | +145% | 0 | 0 | — |
case-22 | pass→pass | 27,700 | 45,691 | +65% | 1 | 1 | 0% | 1,831 | 4,205 | +130% | 0 | 0 | — |
case-23 | fail→pass | 37,303 | 39,909 | +7% | 1 | 1 | 0% | 2,148 | 3,712 | +73% | 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. 23 cases were attempted, and 21 counted toward the lift figure. The other 2 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 +39 percentage points is the difference between those two pass rates over the 21 comparable cases. 2 cases got worse with the skill loaded, and they are 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.