Skip to content
Core Feature

AI Code Review

Automated AI code reviews with 4-phase analysis. HTML and JSON reports, inline PR comments, CI integration. GDPR-compliant, EU-hosted.

🔍

Multi-Perspective Review

Architecture, security, and performance analysis in one structured review.

📊

HTML & JSON Reports

Rich reports for sharing, archiving, and tracking.

CI Integration

`code-review-ci` command for automated pipelines.

📁

Staged & Full Review

Review staged changes or the entire codebase.

🔀

Diff-Base Review

Compare against any branch or commit.

Pre-Checks

Automated lint and type-check before the review runs.

Lurus Code 3-phase AI code review results with severity ratings and fix recommendations

The 4-Phase Review Workflow

Every code review runs through four sequential phases. The result is a structured report with actionable findings — not just a list of style complaints.

1/4

Discovery

Identify which files changed (git diff, staged, or full project). Build a dependency map to understand the blast radius of each change.

2/4

Analysis

Multi-perspective deep dive: code quality, architecture alignment, performance anti-patterns, security issues, test coverage gaps, and documentation completeness.

3/4

Verification

Each finding is cross-referenced against the full codebase. False positives are removed. Remaining findings are ranked by severity and impact.

4/4

Suggestions

For every confirmed finding, the reviewer generates a concrete improvement suggestion — with a code example where applicable.

Output Formats

Choose how you want to consume review results — from a visual HTML dashboard to machine-readable JSON for custom tooling.

🌐

HTML Report

A self-contained visual dashboard with findings grouped by severity, file, and category. Share it with your team or archive it for audits.

📄

JSON Report

Structured findings with file path, line number, severity, category, and suggestion. Ideal for custom dashboards or integrating with your own tooling.

💬

PR Comments

Findings posted as inline review comments on the exact lines in GitHub or GitLab. The AI can also submit an overall verdict: APPROVE or REQUEST_CHANGES.

📟

Terminal Text

Human-readable summary streamed to stdout. Perfect for the interactive /review chat command.

CI/CD Integration

Integrate AI reviews into your pull request workflow with two lines of GitHub Actions config.

github-actions.yml
- name: AI Code Review
  run: lurus code-review-ci --pr-comments --verdict --fail-on high
  env:
    GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

Verdicts

APPROVE No findings at or above the configured severity — PR is automatically approved.
REQUEST_CHANGES Blocking findings found — PR review is submitted with required changes.
COMMENT Findings below threshold — posted as comments without a blocking verdict.

Review Scope Options

Flag Scope Best For
--diff (default) Changed files since last commit Pull requests and feature branches
--staged Only staged (git add) changes Pre-commit hook integration
--full Entire project Initial audit or quarterly review
--diff-base main Comparison against a branch or ref Long-running feature branches
🇪🇺 GDPR & Data Privacy

GDPR & Source Code Privacy

Your source code is processed exclusively on EU-hosted servers and discarded immediately after the review. This makes Lurus Code suitable for regulated industries where code confidentiality is a contractual requirement.

  • Code processed on EU servers only (Germany, France)
  • Zero code retention — discarded after review completes
  • DPA (Art. 28 GDPR) available for paid plans
  • Suitable for fintech, medtech, and government software teams

Frequently Asked Questions

What is AI code review?
AI code review uses machine learning to automatically analyze source code for bugs, security vulnerabilities, style issues, and logic errors. Unlike traditional linters, AI review understands context and intent, catching issues that rule-based tools miss.
How does AI code review compare to manual peer review?
AI code review catches different issues than humans. It excels at finding security vulnerabilities, OWASP Top 10 issues, and pattern violations consistently. Manual review remains better for architecture decisions and business logic validation. The best approach combines both.
Does Lurus Code review my code on EU servers?
Yes. All code review analysis happens exclusively on EU data centers with full GDPR compliance. Your source code never leaves the European Union, and no data is used for AI model training.
What output formats does the code review support?
Lurus Code generates review reports in Markdown (inline), HTML (shareable), JSON (CI/CD integration), and GitHub PR comments. All formats include severity ratings, file locations, and fix suggestions.
Can I integrate AI code review into my CI/CD pipeline?
Yes. The code-review-ci CLI command runs headless reviews in any CI pipeline. It returns standard exit codes, supports GitHub Actions output variables, and can block merges based on review verdicts (approve, request-changes, or comment).

Automate code quality

Integrate AI reviews into your workflow, from the IDE to your CI pipeline.

Get started