Code Review Agent

Senior-level code review that understands your architecture. Custom review agents, full codebase context, and reviews that adapt to your team's standards.

CONFIGURATION

// mesa.config.ts
export default {
  review: {
    agents: [
      {
        name: "security",
        focus: ["auth", "api", "sql"],
        rules: ["no-secrets", "validate-input", "safe-queries"]
      },
      {
        name: "architecture",
        context: "full-codebase",
        focus: ["patterns", "dependencies", "modularity"]
      },
      {
        name: "performance",
        focus: ["loops", "queries", "memory"],
        rules: ["no-n-plus-one", "optimize-renders"]
      }
    ],
    triggers: ["pull_request", "push_to_main"]
  }
}

Fig. 2. Multi-agent review configuration.

Define multiple specialized review agents, each with their own focus areas and rules. Security, architecture, performance—or create custom agents for your team's needs.

Reviews trigger automatically on pull requests, or run on-demand for any commit.

Review Docs Try It Free

EXAMPLE OUTPUT

# AI Review Output

## Security Agent

### HIGH: SQL Injection Risk
`src/api/users.ts:42`
Using string interpolation in SQL query. Use parameterized queries instead.

```diff
- const query = `SELECT * FROM users WHERE id = ${userId}`;
+ const query = 'SELECT * FROM users WHERE id = ?';
+ const result = await db.query(query, [userId]);
```

### MEDIUM: Missing Input Validation
`src/api/users.ts:38`
User input passed directly to database without validation.

## Architecture Agent

### INFO: Consider extracting shared logic
Functions `getUser` and `getAdmin` share 80% of their code.
Consider a shared `getUserByRole` function.

Reviews include severity levels, file locations, explanations, and suggested fixes. Each agent provides feedback in its area of expertise.

Feedback appears inline on GitHub PRs, or via the Mesa dashboard and API.

CAPABILITIES

Custom Review Agents

Define agents specialized for security, architecture, performance, or any domain. Each agent can have its own focus areas and rules.

Full Codebase Context

Agents understand your entire codebase, not just the diff. They catch issues that span multiple files and understand your patterns.

Native GitHub Integration

Reviews appear as inline comments on your PRs. No context switching, no extra tools to check.

Customizable Rules

Define rules in plain English or use built-in rule sets. Rules can reference your documentation and coding standards.

Steerable Feedback

Mark feedback as helpful or not. The agent learns your team's preferences and adjusts its review style over time.

API Access

Trigger reviews programmatically via API. Integrate into your CI/CD pipeline or custom tooling.

PRICING

Only pay for the tokens you use, at cost.

StartupFree

100k lines reviewed / month
10 custom agents
3 GitHub repositories

Get started →

TeamPay-per-token

Unlimited repositories
Unlimited custom agents
Premium models

Get started →

EnterpriseCustom

Self-hosting available
Bring your own API keys
SOC 2 Type II compliant

Get in touch →

Ready to level up your code reviews?

Start with our free tier. No credit card required.

Read Documentation Get Started Free