Comprehensive PR Review

Run a thorough pull request review using multiple specialized agents, each focusing on a different aspect of code quality. For a faster, cheaper review, use /pr-review-quick instead.

When This Skill Applies

• User asks for comprehensive PR review
• User says "/pr-review"
• Before creating or merging a PR

Arguments

Optional arguments to filter which reviews to run:

• comments - Analyze code comment accuracy and maintainability
• tests - Review test coverage quality and completeness
• errors - Check error handling for silent failures
• types - Analyze type design and invariants
• code - General code review for bugs and quality
• simplify - Simplify code for clarity and maintainability
• security - Check for security vulnerabilities
• all - Run all applicable reviews (default)
• parallel - Launch all agents in parallel (faster)

Examples:

/pr-review                    # Full review, sequential
/pr-review tests errors       # Only test coverage and error handling
/pr-review all parallel       # Full review, parallel execution
/pr-review simplify           # Just code simplification

Review Workflow

1. Gather Context

# Get changed files
git diff --name-only main...HEAD 2>/dev/null || git diff --name-only HEAD~5

# Get the actual diff (CRITICAL: agents must only review lines in this diff)
git diff main...HEAD 2>/dev/null || git diff HEAD~5

2. Read Project Guidelines

Find and read relevant CLAUDE.md files:

# Root CLAUDE.md
cat CLAUDE.md 2>/dev/null || true

# CLAUDE.md in directories with changed files
# For each changed directory, check for CLAUDE.md

Extract relevant guidelines that apply to code review (not all instructions apply - focus on code style, patterns, and conventions).

3. Launch Review Agents

CRITICAL INSTRUCTION FOR ALL AGENTS:

Every agent MUST receive these instructions:

IMPORTANT: Only report issues on lines that were ADDED or MODIFIED in this diff.
Do NOT report:
- Pre-existing issues in unchanged code
- Issues on lines that were not modified
- Problems that existed before this change

The diff shows + for added lines and - for removed lines. Only flag issues on + lines.

Sequential (default): Run one at a time for easier action

Parallel (with parallel arg): Launch all at once for speed

For each agent, use the Task tool with this template:

Task (subagent_type: general-purpose): "[AGENT_NAME] REVIEW

CRITICAL: Only report issues on lines ADDED or MODIFIED in this diff (+ lines).
Do NOT report pre-existing issues or problems in unchanged code.

Project guidelines from CLAUDE.md:
[relevant guidelines]

Diff to review:
[paste diff]

[Agent-specific instructions]

For each issue found, provide:
1. File and line number
2. Issue description
3. Confidence score (0-100):
   - 0-25: Likely false positive, doesn't hold up to scrutiny
   - 26-50: Might be real, but could be intentional or a nitpick
   - 51-75: Likely real issue, but may not happen often in practice
   - 76-90: Verified real issue that will impact functionality
   - 91-100: Definitely a bug/problem, confirmed with evidence

Output format:
[CONFIDENCE: XX] file:line - description"

4. Filter Results

After collecting all agent responses:

Only report issues with confidence ≥ 75%.

Filter out:

• Pre-existing issues (not in the diff)
• Issues a linter/typechecker would catch (assume CI runs these)
• Pedantic nitpicks a senior engineer wouldn't flag
• Style issues not explicitly in CLAUDE.md
• Issues on lines the user didn't modify
• Intentional changes related to the broader feature

5. Aggregate Results

Create summary with only high-confidence issues:

# PR Review Summary

**Files reviewed:** [count] **Issues found:** [count] (filtered from [total] raw findings)

## Critical Issues (confidence ≥ 91)

- [file:line] Issue description

## Important Issues (confidence 75-90)

- [file:line] Issue description

## Passed Checks

- [list of agents that found no issues]

## Strengths

- [what's done well in this change]

If no issues pass the confidence threshold: "No significant issues found."

Available Review Agents

code-reviewer

Focus: General code review for project guidelines

• Checks CLAUDE.md compliance
• Detects bugs and logic errors
• Reviews code quality issues

security-reviewer

Focus: Security vulnerabilities

• SQL injection, XSS
• Hardcoded secrets/credentials
• Missing input validation

silent-failure-hunter

Focus: Error handling

• Silent failures in catch blocks
• Inadequate error handling
• Missing error logging

pr-test-analyzer

Focus: Test coverage

• Missing tests for new functionality
• Edge cases not covered
• Test quality issues

type-design-analyzer

Focus: Type design (TypeScript/typed languages)

• Type safety issues
• Missing or overly broad types
• Invariant violations

comment-analyzer

Focus: Documentation accuracy

• Comments that don't match code
• Outdated documentation
• Misleading comments

code-simplifier

Focus: Code simplification (run last)

• Overly complex code
• Unnecessary abstractions
• Opportunities to simplify

False Positive Examples

Train agents to ignore these:

• Something that looks like a bug but isn't
• Issues a linter would catch (imports, formatting)
• General quality issues not in CLAUDE.md
• Issues silenced by lint-ignore comments
• Intentional functionality changes
• Real issues on lines NOT modified in the diff