Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Claude Code Action entegrasyonu - PR review ve issue-to-fix workflow kurulumu
.claude/skills/github-actions-integration/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 85% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 75% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 58% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 3% | 0% |
anthropics/claude-code-action@v1 GitHub Actions action'i. PR'lari otomatik review eder, issue'lardan otomatik fix olusturur. vibecosystem ile entegre calisir.
Kaynak: https://github.com/anthropics/claude-code-action
Repository Settings > Secrets and variables > Actions > New repository secret:
Name: ANTHROPIC_API_KEY
Value: sk-ant-... (Anthropic API key)Dosya: .github/workflows/claude-review.yml
yamlname: Claude Code Review on: pull_request: types: [opened, synchronize, reopened] permissions: contents: read pull-requests: write issues: write jobs: claude-review: runs-on: ubuntu-latest steps: - uses: actions/checkout@v4 with: fetch-depth: 0 - name: Claude Code Review uses: anthropics/claude-code-action@v1 with: anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }} review_comment_prefix: "vibecosystem-review" direct_prompt: | Review this PR for: 1. Code quality and patterns 2. Security vulnerabilities (OWASP Top 10) 3. Performance issues 4. Test coverage gaps 5. Breaking changes Use vibecosystem's code-reviewer agent standards. Provide VERDICT: PASS | WARN | FAIL
Ne yapar: PR acildiginda, guncellediginde veya yeniden acildiginda otomatik review yapar. Review sonucunu PR comment olarak yazar.
Dosya: .github/workflows/claude-fix.yml
yamlname: Claude Code Fix on: issues: types: [labeled] permissions: contents: write pull-requests: write issues: write jobs: claude-fix: if: contains(github.event.label.name, 'claude-fix') runs-on: ubuntu-latest steps: - uses: actions/checkout@v4 - name: Claude Code Fix uses: anthropics/claude-code-action@v1 with: anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }} direct_prompt: | Fix the issue described below. Follow TDD workflow: write test first, then implement fix. Create a PR with the fix. Issue: ${{ github.event.issue.title }} Body: ${{ github.event.issue.body }}
Ne yapar: Issue'ya claude-fix label'i eklendiginde otomatik fix olusturur ve PR acar.
claude-fix label'i ekleSadece belirli dosya tiplerini review et:
yamlon: pull_request: types: [opened, synchronize, reopened] paths: - '**.ts' - '**.tsx' - '**.js' - '**.jsx'
vibecosystem agent standartlarina gore prompt'u genislet:
yamldirect_prompt: | Review this PR using vibecosystem standards: ## Code Quality - Immutability: no mutations, use spread operator - File size: max 800 lines per file - Function size: max 50 lines - Nesting: max 4 levels - No console.log in production code - No hardcoded values ## Security (OWASP Top 10) - Input validation (use zod) - SQL injection prevention - XSS prevention - No hardcoded credentials - Rate limiting check ## Testing - Test coverage >= 80% - Edge cases covered - Error paths tested VERDICT: PASS | WARN | FAIL
yamldirect_prompt: | Fix the issue below. Rules: - TDD: write failing test first, then implement - Defensive code: try/catch, null checks, input validation - No mutations, use immutable patterns - Follow repository's existing code style - Run tests before creating PR Issue: ${{ github.event.issue.title }} Body: ${{ github.event.issue.body }}
Farkli label'lar icin farkli davranislar:
yamljobs: claude-fix: if: contains(github.event.label.name, 'claude-fix') # ... standard fix claude-refactor: if: contains(github.event.label.name, 'claude-refactor') runs-on: ubuntu-latest steps: - uses: actions/checkout@v4 - uses: anthropics/claude-code-action@v1 with: anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }} direct_prompt: | Refactor the code described in this issue. Maintain all existing tests. Do not change external behavior. Issue: ${{ github.event.issue.title }} Body: ${{ github.event.issue.body }}
Repository'de CLAUDE.md varsa claude-code-action onu otomatik okur. vibecosystem kurallarini CLAUDE.md'ye ekle:
markdown# Project Rules - Follow vibecosystem coding standards - Use TDD workflow for all changes - Security review required for auth/data code - VERDICT format: PASS | WARN | FAIL
Review workflow'u vibecosystem'in code-reviewer agent standartlarini kullanir:
Fix workflow'u TDD workflow'unu kullanir:
| Sorun | Cozum | |-------|-------| | Review calismadi | ANTHROPIC_API_KEY secret ayarli mi? | | Permission hatasi | Workflow permissions blogu dogru mu? | | Fix PR acilmadi | claude-fix label dogru yazildi mi? | | Review cok uzun suruyor | fetch-depth: 1 ile sadece son commit'i al | | Rate limit | API key plan limitini kontrol et |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 10,353 | 9,588 | -7% | 1 | 1 | 0% | 2,021 | 3,737 | +85% | 0 | 0 | — |
case-02 | fail→fail | 14,600 | 7,821 | -46% | 1 | 1 | 0% | 2,325 | 3,303 | +42% | 0 | 0 | — |
case-03 | fail→pass | 17,240 | 10,154 | -41% | 1 | 1 | 0% | 3,098 | 3,724 | +20% | 0 | 0 | — |
case-04 | pass→pass | 6,807 | 2,555 | -62% | 1 | 1 | 0% | 1,238 | 2,165 | +75% | 0 | 0 | — |
case-05 | fail→pass | 8,255 | 4,379 | -47% | 1 | 1 | 0% | 1,453 | 2,549 | +75% | 0 | 0 | — |
case-06 | pass→pass | 5,856 | 4,215 | -28% | 1 | 1 | 0% | 1,077 | 2,651 | +146% | 0 | 0 | — |
case-07 | fail→pass | 6,886 | 1,898 | -72% | 1 | 1 | 0% | 1,317 | 2,076 | +58% | 0 | 0 | — |
case-08 | fail→pass | 11,570 | 2,749 | -76% | 1 | 1 | 0% | 2,173 | 2,232 | +3% | 0 | 0 | — |
case-09 | pass→pass | 5,648 | 2,436 | -57% | 1 | 1 | 0% | 895 | 2,121 | +137% | 0 | 0 | — |
case-10 | fail→fail | 11,955 | 7,760 | -35% | 1 | 1 | 0% | 2,259 | 3,357 | +49% | 0 | 0 | — |
case-11 | pass→pass | 6,514 | 5,088 | -22% | 1 | 1 | 0% | 1,220 | 2,439 | +100% | 0 | 0 | — |
case-12 | pass→pass | 10,596 | 2,536 | -76% | 1 | 1 | 0% | 1,779 | 2,203 | +24% | 0 | 0 | — |
case-13 | pass→pass | 5,262 | 10,495 | +99% | 1 | 1 | 0% | 934 | 3,364 | +260% | 0 | 0 | — |
case-14 | pass→pass | 7,594 | 6,931 | -9% | 1 | 1 | 0% | 1,376 | 3,079 | +124% | 0 | 0 | — |
case-15 | pass→pass | 7,274 | 1,689 | -77% | 1 | 1 | 0% | 1,074 | 2,023 | +88% | 0 | 0 | — |
case-16 | fail→pass | 9,690 | 3,261 | -66% | 1 | 1 | 0% | 1,664 | 2,304 | +38% | 0 | 0 | — |
case-17 | pass→pass | 5,174 | 2,912 | -44% | 1 | 1 | 0% | 999 | 2,319 | +132% | 0 | 0 | — |
case-18 | pass→pass | 12,993 | 13,207 | +2% | 1 | 1 | 0% | 2,328 | 4,193 | +80% | 0 | 0 | — |
case-19 | pass→pass | 9,287 | 4,142 | -55% | 1 | 1 | 0% | 1,514 | 2,533 | +67% | 0 | 0 | — |
case-20 | pass→pass | 7,894 | 5,337 | -32% | 1 | 1 | 0% | 1,655 | 2,857 | +73% | 0 | 0 | — |
case-21 | pass→pass | 7,731 | 5,740 | -26% | 1 | 1 | 0% | 1,503 | 2,961 | +97% | 0 | 0 | — |
case-22 | pass→pass | 9,117 | 10,681 | +17% | 1 | 1 | 0% | 1,910 | 3,560 | +86% | 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 +27 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 7/29/2026 | +27% |
Other measured skills in the registry, with their headline benchmark lift.