AI Code Review Command

Performs AI-powered code review using the embedded ai-code-agent-mcp server with Codex and/or Gemini CLI tools.

Usage

/ai-review [scope] [options]
/cr [scope] [options]

Arguments

Scope (what to review)

• file <path> - Review a specific file
• diff - Review uncommitted git changes
• staged - Review staged changes only
• pr [number] - Review pull request changes
• (none) - Review current context/conversation code

Options

Model Selection

• --model=codex - Use only OpenAI Codex
• --model=gemini - Use only Google Gemini
• --model=combined - Use both with result aggregation (default)

Analysis Focus

• --focus=security - Security-focused review
• --focus=performance - Performance-focused review
• --focus=bugs - Bug detection focus
• --focus=style - Code style and best practices
• --focus=all - Comprehensive review (default)

Severity Filter

• --severity=critical - Show only critical issues
• --severity=high - Show high and critical issues
• --severity=all - Show all issues (default)

Context Preset

• --preset=react-web - React web application
• --preset=nodejs-api - Node.js API server
• --preset=mcp-server - MCP server project
• --preset=cli-tool - CLI tool
• --preset=library - Reusable library
• --preset=auto - Auto-detect from code (default)

Output Options

• --format=markdown - Markdown output (default)
• --format=json - JSON structured output
• --verbose - Include detailed findings

Examples

# Review current context with both models
/cr

# Review specific file with Codex only
/cr file src/api/auth.ts --model=codex

# Security-focused review of git diff
/cr diff --focus=security

# Review staged changes with high severity filter
/cr staged --severity=high

# Combined review with auto-detection
/cr file lib/parser.js --model=combined --preset=auto

# Verbose security review
/cr file .env.example --focus=security --verbose

Prerequisites

This command requires the ai-code-agent-mcp MCP server and at least one CLI tool:

Codex CLI

# Install
npm install -g @openai/codex

# Authenticate
codex auth

Gemini CLI

# Install
npm install -g @google/gemini-cli

# Authenticate
gemini auth login

Output Example

## AI Code Review Results

**Analysis ID**: combined-1735300000000
**Models Used**: codex, gemini
**Duration**: 3240ms

### Overall Assessment

The code demonstrates good structure but has potential security issues in input validation and a performance concern with database queries.

### Summary

| Severity | Count |
|----------|-------|
| Critical | 1 |
| High | 2 |
| Medium | 4 |
| Low | 3 |
| **Total** | **10** |

### Findings

#### Critical Issues

1. **SQL Injection Vulnerability** (Line 42)
   - **Description**: User input directly concatenated into SQL query
   - **Suggestion**: Use parameterized queries
   - **Confidence**: High (both models agree)

...

Related Commands

• /r - Quick Claude-only code review
• /security-scan - Security-focused scan with security-scanner skill
• /context-setup - Configure MCP servers including ai-code-agent-mcp

Troubleshooting

"CLI not found" Error

Ensure Codex or Gemini CLI is installed and in PATH:

which codex  # or: where codex (Windows)
which gemini

"Authentication failed" Error

Re-authenticate with the respective CLI:

codex auth
gemini auth login

Slow Analysis

• Check network connection
• Consider using --model=codex or --model=gemini for faster single-model review
• Large files are automatically truncated; consider splitting

Cache Issues

Analysis results are cached for 1 hour. To force fresh analysis:

# Clear cache (if needed)
rm -rf ~/.claude/claude-vibe/data/ai-code-agent.db