surshar / claude-code
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.
Built by Anthropic · github.com/anthropics
| 1 | --- |
| 2 | description: "Comprehensive PR review using specialized agents" |
| 3 | argument-hint: "[review-aspects]" |
| 4 | allowed-tools: ["Bash", "Glob", "Grep", "Read", "Task"] |
| 5 | --- |
| 6 | |
| 7 | # Comprehensive PR Review |
| 8 | |
| 9 | Run a comprehensive pull request review using multiple specialized agents, each focusing on a different aspect of code quality. |
| 10 | |
| 11 | **Review Aspects (optional):** "$ARGUMENTS" |
| 12 | |
| 13 | ## Review Workflow: |
| 14 | |
| 15 | 1. **Determine Review Scope** |
| 16 | - Check git status to identify changed files |
| 17 | - Parse arguments to see if user requested specific review aspects |
| 18 | - Default: Run all applicable reviews |
| 19 | |
| 20 | 2. **Available Review Aspects:** |
| 21 | |
| 22 | - **comments** - Analyze code comment accuracy and maintainability |
| 23 | - **tests** - Review test coverage quality and completeness |
| 24 | - **errors** - Check error handling for silent failures |
| 25 | - **types** - Analyze type design and invariants (if new types added) |
| 26 | - **code** - General code review for project guidelines |
| 27 | - **simplify** - Simplify code for clarity and maintainability |
| 28 | - **all** - Run all applicable reviews (default) |
| 29 | |
| 30 | 3. **Identify Changed Files** |
| 31 | - Run `git diff --name-only` to see modified files |
| 32 | - Check if PR already exists: `gh pr view` |
| 33 | - Identify file types and what reviews apply |
| 34 | |
| 35 | 4. **Determine Applicable Reviews** |
| 36 | |
| 37 | Based on changes: |
| 38 | - **Always applicable**: code-reviewer (general quality) |
| 39 | - **If test files changed**: pr-test-analyzer |
| 40 | - **If comments/docs added**: comment-analyzer |
| 41 | - **If error handling changed**: silent-failure-hunter |
| 42 | - **If types added/modified**: type-design-analyzer |
| 43 | - **After passing review**: code-simplifier (polish and refine) |
| 44 | |
| 45 | 5. **Launch Review Agents** |
| 46 | |
| 47 | **Sequential approach** (one at a time): |
| 48 | - Easier to understand and act on |
| 49 | - Each report is complete before next |
| 50 | - Good for interactive review |
| 51 | |
| 52 | **Parallel approach** (user can request): |
| 53 | - Launch all agents simultaneously |
| 54 | - Faster for comprehensive review |
| 55 | - Results come back together |
| 56 | |
| 57 | 6. **Aggregate Results** |
| 58 | |
| 59 | After agents complete, summarize: |
| 60 | - **Critical Issues** (must fix before merge) |
| 61 | - **Important Issues** (should fix) |
| 62 | - **Suggestions** (nice to have) |
| 63 | - **Positive Observations** (what's good) |
| 64 | |
| 65 | 7. **Provide Action Plan** |
| 66 | |
| 67 | Organize findings: |
| 68 | ```markdown |
| 69 | # PR Review Summary |
| 70 | |
| 71 | ## Critical Issues (X found) |
| 72 | - [agent-name]: Issue description [file:line] |
| 73 | |
| 74 | ## Important Issues (X found) |
| 75 | - [agent-name]: Issue description [file:line] |
| 76 | |
| 77 | ## Suggestions (X found) |
| 78 | - [agent-name]: Suggestion [file:line] |
| 79 | |
| 80 | ## Strengths |
| 81 | - What's well-done in this PR |
| 82 | |
| 83 | ## Recommended Action |
| 84 | 1. Fix critical issues first |
| 85 | 2. Address important issues |
| 86 | 3. Consider suggestions |
| 87 | 4. Re-run review after fixes |
| 88 | ``` |
| 89 | |
| 90 | ## Usage Examples: |
| 91 | |
| 92 | **Full review (default):** |
| 93 | ``` |
| 94 | /pr-review-toolkit:review-pr |
| 95 | ``` |
| 96 | |
| 97 | **Specific aspects:** |
| 98 | ``` |
| 99 | /pr-review-toolkit:review-pr tests errors |
| 100 | # Reviews only test coverage and error handling |
| 101 | |
| 102 | /pr-review-toolkit:review-pr comments |
| 103 | # Reviews only code comments |
| 104 | |
| 105 | /pr-review-toolkit:review-pr simplify |
| 106 | # Simplifies code after passing review |
| 107 | ``` |
| 108 | |
| 109 | **Parallel review:** |
| 110 | ``` |
| 111 | /pr-review-toolkit:review-pr all parallel |
| 112 | # Launches all agents in parallel |
| 113 | ``` |
| 114 | |
| 115 | ## Agent Descriptions: |
| 116 | |
| 117 | **comment-analyzer**: |
| 118 | - Verifies comment accuracy vs code |
| 119 | - Identifies comment rot |
| 120 | - Checks documentation completeness |
| 121 | |
| 122 | **pr-test-analyzer**: |
| 123 | - Reviews behavioral test coverage |
| 124 | - Identifies critical gaps |
| 125 | - Evaluates test quality |
| 126 | |
| 127 | **silent-failure-hunter**: |
| 128 | - Finds silent failures |
| 129 | - Reviews catch blocks |
| 130 | - Checks error logging |
| 131 | |
| 132 | **type-design-analyzer**: |
| 133 | - Analyzes type encapsulation |
| 134 | - Reviews invariant expression |
| 135 | - Rates type design quality |
| 136 | |
| 137 | **code-reviewer**: |
| 138 | - Checks CLAUDE.md compliance |
| 139 | - Detects bugs and issues |
| 140 | - Reviews general code quality |
| 141 | |
| 142 | **code-simplifier**: |
| 143 | - Simplifies complex code |
| 144 | - Improves clarity and readability |
| 145 | - Applies project standards |
| 146 | - Preserves functionality |
| 147 | |
| 148 | ## Tips: |
| 149 | |
| 150 | - **Run early**: Before creating PR, not after |
| 151 | - **Focus on changes**: Agents analyze git diff by default |
| 152 | - **Address critical first**: Fix high-priority issues before lower priority |
| 153 | - **Re-run after fixes**: Verify issues are resolved |
| 154 | - **Use specific reviews**: Target specific aspects when you know the concern |
| 155 | |
| 156 | ## Workflow Integration: |
| 157 | |
| 158 | **Before committing:** |
| 159 | ``` |
| 160 | 1. Write code |
| 161 | 2. Run: /pr-review-toolkit:review-pr code errors |
| 162 | 3. Fix any critical issues |
| 163 | 4. Commit |
| 164 | ``` |
| 165 | |
| 166 | **Before creating PR:** |
| 167 | ``` |
| 168 | 1. Stage all changes |
| 169 | 2. Run: /pr-review-toolkit:review-pr all |
| 170 | 3. Address all critical and important issues |
| 171 | 4. Run specific reviews again to verify |
| 172 | 5. Create PR |
| 173 | ``` |
| 174 | |
| 175 | **After PR feedback:** |
| 176 | ``` |
| 177 | 1. Make requested changes |
| 178 | 2. Run targeted reviews based on feedback |
| 179 | 3. Verify issues are resolved |
| 180 | 4. Push updates |
| 181 | ``` |
| 182 | |
| 183 | ## Notes: |
| 184 | |
| 185 | - Agents run autonomously and return detailed reports |
| 186 | - Each agent focuses on its specialty for deep analysis |
| 187 | - Results are actionable with specific file:line references |
| 188 | - Agents use appropriate models for their complexity |
| 189 | - All agents available in `/agents` list |