1# PR Review Toolkit
2 
3A comprehensive collection of specialized agents for thorough pull request review, covering code comments, test coverage, error handling, type design, code quality, and code simplification.
4 
5## Overview
6 
7This plugin bundles 6 expert review agents that each focus on a specific aspect of code quality. Use them individually for targeted reviews or together for comprehensive PR analysis.
8 
9## Agents
10 
11### 1. comment-analyzer
12**Focus**: Code comment accuracy and maintainability
13 
14**Analyzes:**
15- Comment accuracy vs actual code
16- Documentation completeness
17- Comment rot and technical debt
18- Misleading or outdated comments
19 
20**When to use:**
21- After adding documentation
22- Before finalizing PRs with comment changes
23- When reviewing existing comments
24 
25**Triggers:**
26```
27"Check if the comments are accurate"
28"Review the documentation I added"
29"Analyze comments for technical debt"
30```
31 
32### 2. pr-test-analyzer
33**Focus**: Test coverage quality and completeness
34 
35**Analyzes:**
36- Behavioral vs line coverage
37- Critical gaps in test coverage
38- Test quality and resilience
39- Edge cases and error conditions
40 
41**When to use:**
42- After creating a PR
43- When adding new functionality
44- To verify test thoroughness
45 
46**Triggers:**
47```
48"Check if the tests are thorough"
49"Review test coverage for this PR"
50"Are there any critical test gaps?"
51```
52 
53### 3. silent-failure-hunter
54**Focus**: Error handling and silent failures
55 
56**Analyzes:**
57- Silent failures in catch blocks
58- Inadequate error handling
59- Inappropriate fallback behavior
60- Missing error logging
61 
62**When to use:**
63- After implementing error handling
64- When reviewing try/catch blocks
65- Before finalizing PRs with error handling
66 
67**Triggers:**
68```
69"Review the error handling"
70"Check for silent failures"
71"Analyze catch blocks in this PR"
72```
73 
74### 4. type-design-analyzer
75**Focus**: Type design quality and invariants
76 
77**Analyzes:**
78- Type encapsulation (rated 1-10)
79- Invariant expression (rated 1-10)
80- Type usefulness (rated 1-10)
81- Invariant enforcement (rated 1-10)
82 
83**When to use:**
84- When introducing new types
85- During PR creation with data models
86- When refactoring type designs
87 
88**Triggers:**
89```
90"Review the UserAccount type design"
91"Analyze type design in this PR"
92"Check if this type has strong invariants"
93```
94 
95### 5. code-reviewer
96**Focus**: General code review for project guidelines
97 
98**Analyzes:**
99- CLAUDE.md compliance
100- Style violations
101- Bug detection
102- Code quality issues
103 
104**When to use:**
105- After writing or modifying code
106- Before committing changes
107- Before creating pull requests
108 
109**Triggers:**
110```
111"Review my recent changes"
112"Check if everything looks good"
113"Review this code before I commit"
114```
115 
116### 6. code-simplifier
117**Focus**: Code simplification and refactoring
118 
119**Analyzes:**
120- Code clarity and readability
121- Unnecessary complexity and nesting
122- Redundant code and abstractions
123- Consistency with project standards
124- Overly compact or clever code
125 
126**When to use:**
127- After writing or modifying code
128- After passing code review
129- When code works but feels complex
130 
131**Triggers:**
132```
133"Simplify this code"
134"Make this clearer"
135"Refine this implementation"
136```
137 
138**Note**: This agent preserves functionality while improving code structure and maintainability.
139 
140## Usage Patterns
141 
142### Individual Agent Usage
143 
144Simply ask questions that match an agent's focus area, and Claude will automatically trigger the appropriate agent:
145 
146```
147"Can you check if the tests cover all edge cases?"
148→ Triggers pr-test-analyzer
149 
150"Review the error handling in the API client"
151→ Triggers silent-failure-hunter
152 
153"I've added documentation - is it accurate?"
154→ Triggers comment-analyzer
155```
156 
157### Comprehensive PR Review
158 
159For thorough PR review, ask for multiple aspects:
160 
161```
162"I'm ready to create this PR. Please:
1631. Review test coverage
1642. Check for silent failures
1653. Verify code comments are accurate
1664. Review any new types
1675. General code review"
168```
169 
170This will trigger all relevant agents to analyze different aspects of your PR.
171 
172### Proactive Review
173 
174Claude may proactively use these agents based on context:
175 
176- **After writing code** → code-reviewer
177- **After adding docs** → comment-analyzer
178- **Before creating PR** → Multiple agents as appropriate
179- **After adding types** → type-design-analyzer
180 
181## Installation
182 
183Install from your personal marketplace:
184 
185```bash
186/plugins
187# Find "pr-review-toolkit"
188# Install
189```
190 
191Or add manually to settings if needed.
192 
193## Agent Details
194 
195### Confidence Scoring
196 
197Agents provide confidence scores for their findings:
198 
199**comment-analyzer**: Identifies issues with high confidence in accuracy checks
200 
201**pr-test-analyzer**: Rates test gaps 1-10 (10 = critical, must add)
202 
203**silent-failure-hunter**: Flags severity of error handling issues
204 
205**type-design-analyzer**: Rates 4 dimensions on 1-10 scale
206 
207**code-reviewer**: Scores issues 0-100 (91-100 = critical)
208 
209**code-simplifier**: Identifies complexity and suggests simplifications
210 
211### Output Formats
212 
213All agents provide structured, actionable output:
214- Clear issue identification
215- Specific file and line references
216- Explanation of why it's a problem
217- Suggestions for improvement
218- Prioritized by severity
219 
220## Best Practices
221 
222### When to Use Each Agent
223 
224**Before Committing:**
225- code-reviewer (general quality)
226- silent-failure-hunter (if changed error handling)
227 
228**Before Creating PR:**
229- pr-test-analyzer (test coverage check)
230- comment-analyzer (if added/modified comments)
231- type-design-analyzer (if added/modified types)
232- code-reviewer (final sweep)
233 
234**After Passing Review:**
235- code-simplifier (improve clarity and maintainability)
236 
237**During PR Review:**
238- Any agent for specific concerns raised
239- Targeted re-review after fixes
240 
241### Running Multiple Agents
242 
243You can request multiple agents to run in parallel or sequentially:
244 
245**Parallel** (faster):
246```
247"Run pr-test-analyzer and comment-analyzer in parallel"
248```
249 
250**Sequential** (when one informs the other):
251```
252"First review test coverage, then check code quality"
253```
254 
255## Tips
256 
257- **Be specific**: Target specific agents for focused review
258- **Use proactively**: Run before creating PRs, not after
259- **Address critical issues first**: Agents prioritize findings
260- **Iterate**: Run again after fixes to verify
261- **Don't over-use**: Focus on changed code, not entire codebase
262 
263## Troubleshooting
264 
265### Agent Not Triggering
266 
267**Issue**: Asked for review but agent didn't run
268 
269**Solution**:
270- Be more specific in your request
271- Mention the agent type explicitly
272- Reference the specific concern (e.g., "test coverage")
273 
274### Agent Analyzing Wrong Files
275 
276**Issue**: Agent reviewing too much or wrong files
277 
278**Solution**:
279- Specify which files to focus on
280- Reference the PR number or branch
281- Mention "recent changes" or "git diff"
282 
283## Integration with Workflow
284 
285This plugin works great with:
286- **build-validator**: Run build/tests before review
287- **Project-specific agents**: Combine with your custom agents
288 
289**Recommended workflow:**
2901. Write code → **code-reviewer**
2912. Fix issues → **silent-failure-hunter** (if error handling)
2923. Add tests → **pr-test-analyzer**
2934. Document → **comment-analyzer**
2945. Review passes → **code-simplifier** (polish)
2956. Create PR
296 
297## Contributing
298 
299Found issues or have suggestions? These agents are maintained in:
300- User agents: `~/.claude/agents/`
301- Project agents: `.claude/agents/` in claude-cli-internal
302 
303## License
304 
305MIT
306 
307## Author
308 
309Daisy ([email protected])
310 
311---
312 
313**Quick Start**: Just ask for review and the right agent will trigger automatically!
314