MCP Code Review Server – AI code review in your editor

Code review as an MCP server. Connect it to Claude Code, Cursor, or any MCP-compatible AI assistant.

— Review any source code snippet for bugs, security, performance, and stylereview_code

— Review a git diff for potential issues before mergingreview_diff

— Review a local file by pathreview_file

Powered by the same methodology as our Code Review Agent: OWASP Top 10 scanning, N+1 query detection, race condition analysis, and structured output with severity ratings.

claude mcp add code-review -- uvx aicraft-code-review

Or add to your ~/.cursor/mcp.json

or claude_desktop_config.json

:

{
  "mcpServers": {
    "code-review": {
      "command": "uvx",
      "args": ["mcp-code-review"]
    }
  }
}
pip install aicraft-code-review
python -m mcp_code_review

Once connected, ask your AI assistant:

"Review this Python code for security issues: [paste code]" "Review this diff before I commit: [paste diff]" "Review this file: /path/to/file.py"

The AI will call the MCP server and return structured results.

## Review Results

### 🔴 Critical (1)
| Line | Issue | Category | Fix |
|------|-------|----------|-----|
| 42 | SQL injection via f-string | Security | Use parameterized queries |

### 🟠 High (2)
| Line | Issue | Category | Fix |
|------|-------|----------|-----|
| 15 | Unvalidated user input | Security | Add input validation |
| 78 | N+1 query in loop | Performance | Add select_related |

### Summary
- **Critical**: 1 — must fix
- **High**: 2 — should fix
- **Medium**: 0
- **Info**: 0
git clone https://github.com/GoodJobwilliam/aicraft
cd aicraft
pip install -e ".[dev]"
python -m mcp_code_review  # Start server

• Python 3.11+
• An MCP-compatible client (Claude Code, Cursor, etc.)

MIT