review（审查）

User request: $ARGUMENTS

Orchestrate comprehensive code review by running specialized review skills and consolidating findings.

Flags:

• --autonomous → no user prompts, run all applicable reviews, return consolidated report
• --skip <types> → skip specific review types (comma-separated: bugs, types, maintainability, simplicity, testability, coverage, docs, agents-md)
• --only <types> → run only specific review types

Output: Consolidated review report to user.

Phase 1: Setup

1.1 Determine Scope

Priority order for scope:

1. User specifies files/directories in $ARGUMENTS → use those
2. Otherwise → diff against origin/main or origin/master
3. No changes found → ask user what to review

1.2 Select Reviews

Determine which reviews to run based on scope and arguments:

Detection logic:

• TypeScript: tsconfig.json exists
• Typed Python: pyproject.toml with mypy config OR py.typed marker
• Test files: *.test.*, *.spec.*, __tests__/, tests/ exist
• AGENTS.md: file exists in project root or parent directories

1.3 Create Todo List

- [ ] Run bug review
- [ ] Run type safety review (if applicable)
- [ ] Run maintainability review
- [ ] Run simplicity review
- [ ] Run testability review
- [ ] Run coverage review (if applicable)
- [ ] Run docs review
- [ ] Run AGENTS.md review (if applicable)
- [ ] Consolidate findings

Phase 2: Execute Reviews

2.1 Review Loop

For each selected review type:

1. Mark todo in_progress
2. Execute the review skill
3. Capture findings (severity, location, description, suggested fix)
4. Mark todo completed
5. Continue to next review

2.2 Individual Review Execution

Run each review against the same scope:

Bug Review:

$review-bugs {scope}

Type Safety Review (if TypeScript/typed Python):

$review-type-safety {scope}

Maintainability Review:

$review-maintainability {scope}

Simplicity Review:

$review-simplicity {scope}

Testability Review:

$review-testability {scope}

Coverage Review (if tests exist):

$review-coverage {scope}

Docs Review:

$review-docs {scope}

AGENTS.md Review (if AGENTS.md exists):

$review-agents-md-adherence {scope}

2.3 Capture Findings

From each review, extract issues with:

• Category: bugs | type-safety | maintainability | simplicity | testability | coverage | docs | agents-md
• Severity: Critical | High | Medium | Low
• Location: file:line
• Description: What the issue is
• Suggested Fix: How to resolve

Phase 3: Consolidate

3.1 Merge and Deduplicate

Combine all findings:

1. Group by severity (Critical → High → Medium → Low)
2. Within severity, group by category
3. Deduplicate: same file:line with similar description → merge, keep highest severity

3.2 Generate Report

# Code Review Report

**Scope**: {files/changes reviewed}
**Reviews run**: {list of review types executed}

## Summary

| Category | Critical | High | Medium | Low |
|----------|----------|------|--------|-----|
| Bugs | N | N | N | N |
| Type Safety | N | N | N | N |
| Maintainability | N | N | N | N |
| Simplicity | N | N | N | N |
| Testability | N | N | N | N |
| Coverage | N | N | N | N |
| Docs | N | N | N | N |
| AGENTS.md | N | N | N | N |
| **Total** | N | N | N | N |

## Critical Issues

### [BUG] {Title}
**Location**: `file.ts:line`
**Description**: {issue}
**Suggested Fix**: {fix}

### [TYPE] {Title}
...

## High Issues
{Same format}

## Medium Issues
{Same format}

## Low Issues
{Same format}

## Review Status

| Review | Status | Issues Found |
|--------|--------|--------------|
| Bugs | ✓ Complete | N |
| Type Safety | ✓ Complete | N |
| Maintainability | ✓ Complete | N |
| Simplicity | ✓ Complete | N |
| Testability | ✓ Complete | N |
| Coverage | ✓ Complete | N |
| Docs | ✓ Complete | N |
| AGENTS.md | ✓ Complete | N |

## Recommendations

### Priority Fixes (do first)
1. {Critical/High issue with highest impact}
2. {Second priority}
3. {Third priority}

### Quick Wins (easy fixes)
- {Low-effort issues that can be fixed quickly}

### Deferred (consider later)
- {Low-severity issues that don't block merge}

3.3 Clean Report

If no issues found:

# Code Review Report

**Scope**: {files/changes reviewed}
**Reviews run**: {list}
**Status**: ALL CLEAR

No issues found. Code is ready for merge.

## Review Status

| Review | Status | Issues Found |
|--------|--------|--------------|
| Bugs | ✓ Complete | 0 |
| Type Safety | ✓ Complete | 0 |
...

Phase 4: Present Results

4.1 If --autonomous

Return consolidated report without prompts.

4.2 If Interactive

Present summary and ask:

## Review Complete

Found {N} issues across {M} categories.

Critical: {N} | High: {N} | Medium: {N} | Low: {N}

### Top Priority Fixes
1. {Issue 1}
2. {Issue 2}
3. {Issue 3}

What would you like to do?
Options:
  - Fix issues - run $fix-review-issues (Recommended if Critical/High exist)
  - Show full report - see all findings
  - Done - I'll address manually

Edge Cases

Guidelines

• Comprehensive: Run all applicable reviews unless explicitly skipped
• Consolidated: Single unified report, not multiple separate outputs
• Actionable: Prioritize findings, suggest fix order
• Efficient: Skip inapplicable reviews (no TypeScript = skip type review)
• Autonomous: Support --autonomous for CI/automation use