The Code Review Agent is a built-in capability in Agent mode that automates code review. Specify a scope—an entire project, specific files, a Git diff, or a Pull Request—and the agent scans for defects, security vulnerabilities, performance issues, style problems, readability issues, and best practices violations, then returns a structured report with severity-ranked findings and suggested fixes.
In the chat, use the /code-review command or describe what to review. The agent analyzes the target code and generates a structured review report.
Core capabilities
-
Multi-dimensional review: Detects logical defects, security vulnerabilities (such as SQL injection and XSS), performance bottlenecks, and code style issues.
-
Flexible review scope: Supports scans of entire projects, specific files, Git diffs, and Pull Requests.
-
Actionable findings: Categorizes findings by severity (Error, Warning, Suggestion). Each finding includes the problem location, an explanation, and recommended fixes.
How to use
The Code Review Agent is built into Agent mode and requires no separate configuration. Open the chat panel, switch to Agent mode, and use the /code-review command, specifying the review scope as needed.
For large changes, review code in batches by module to get more detailed feedback. If a change involves specific business logic, briefly explain the context so the agent can provide more relevant suggestions.