Code Reviewer

code-reviewer is a subagent from the VoltAgent/awesome-claude-code-subagents collection (MIT license, 158+ subagents across ten categories per the repository), file categories/04-quality-security/code-reviewer.md.

A subagent is a single markdown file with YAML frontmatter that gives Claude Code a distinct role with its own system prompt and tool selection. code-reviewer's frontmatter, per the source, reads:

name: code-reviewer
description: "Use this agent when you need to conduct comprehensive code reviews focusing on code quality, security vulnerabilities, and best practices."
tools: Read, Write, Edit, Bash, Glob, Grep
model: inherit

The system prompt positions the agent as a senior code reviewer focused on code quality, security vulnerabilities, and optimisation opportunities across multiple programming languages. Among the things it checks: test coverage above 80%, cyclomatic complexity below 10, security vulnerabilities, documentation completeness, code smells, and adherence to design patterns.

Installation per the repository: copy the file into .claude/agents/ (project-level) or ~/.claude/agents/ (user-level); the repository also offers an interactive installer and a plugin-marketplace path.

What sets a subagent apart from a plain prompt

Instead of re-explaining what a code review should look for on every request, the subagent permanently encapsulates those instructions in its own file with a clearly defined role, a restricted tool selection, and its own system prompt. Once code-reviewer is installed, Claude Code can activate the role automatically for matching tasks or invoke it by name, without having to restate the review scope every time. That makes recurring reviews more consistent, since the same criteria — test coverage, cyclomatic complexity, security vulnerabilities — are applied the same way on every invocation.

Who the subagent fits

code-reviewer is particularly useful for solo developers and small teams that don't have a dedicated code-review process with multiple human reviewers but still want a structured check before merging. Since the subagent is language-agnostic, it suits mixed codebases spanning several programming languages. Compared with Anthropic's official Code Review plugin, code-reviewer operates as a single role instead of four parallel specialist agents and isn't tied to GitHub pull requests — it can also be applied locally to any code.

Frequently asked questions

Do I need to use GitHub to use the subagent? No, unlike the official Code Review plugin, code-reviewer isn't tied to pull requests and works on any local code. Can I customize the review criteria? Yes, since it's a plain markdown file, the system prompt can be freely edited after installation. Is the subagent official from Anthropic? No, it comes from VoltAgent's community collection, not from Anthropic itself. Does the subagent work with any Claude model? Per the frontmatter, it inherits the session's active model (model: inherit), so no specific model is hardcoded.

cp categories/04-quality-security/code-reviewer.md ~/.claude/agents/code-reviewer.md