AI ReviewBot

AI-powered code review tool for GitHub and GitLab that provides intelligent feedback with inline suggestions and one-click "Apply" button.

📚 Documentation • 🚀 Quick Start • 🛒 GitHub Marketplace

✨ Features

• 🤖 AI-Powered Analysis — Uses Google Gemini or Mistral AI for deep code understanding
• 💡 Inline Suggestions — Comments directly on code lines with GitHub's "Apply suggestion" button
• 🔒 Security Focus — Identifies vulnerabilities with severity levels (Critical, Warning, Info)
• 🌍 Multi-Language — Responds in your PR/MR language (adaptive mode)
• ✨ Good Practices — Highlights what you're doing right, not just issues
• 📊 Transparent Metrics — Shows tokens, latency, and estimated cost
• 🦊 GitHub & GitLab — Native support for both platforms

🔍 Project Discovery

AI ReviewBot automatically analyzes your repository before each review:

• Languages & frameworks detected from GitHub/GitLab API
• CI pipeline parsed to understand what's already automated (linters, formatters, type checkers)
• Config files read to understand project conventions
• Attention Zones classify each quality area as ✅ Well Covered, ⚠️ Weakly Covered, or ❌ Not Covered

This means the bot won't duplicate feedback that your CI already catches — no formatting nits when you run Prettier, no type errors when you run mypy. Discovery results are cached via watch-files, so repeated PRs cost 0 extra tokens.

Create .reviewbot.md in your repo root to customize. You can also run ai-review discover owner/repo to preview what the bot sees.

See Discovery documentation for details.

🚀 Quick Start

GitHub Actions (Recommended)

# .github/workflows/ai-review.yml
name: AI Code Review

on:
  pull_request:
    types: [opened, synchronize]

jobs:
  review:
    runs-on: ubuntu-latest
    permissions:
      contents: read
      pull-requests: write

    steps:
      - uses: KonstZiv/ai-code-reviewer@v1
        with:
          github_token: ${{ secrets.GITHUB_TOKEN }}
          google_api_key: ${{ secrets.AI_REVIEWER_GOOGLE_API_KEY }}
          # Or use Mistral:
          # mistral_api_key: ${{ secrets.AI_REVIEWER_MISTRAL_API_KEY }}
          # llm_provider: mistral

GitLab CI

# .gitlab-ci.yml
ai-review:
  image: ghcr.io/konstziv/ai-code-reviewer:1
  script:
    - ai-review
  rules:
    - if: $CI_PIPELINE_SOURCE == "merge_request_event"

Note: Set LLM API key (AI_REVIEWER_GOOGLE_API_KEY or AI_REVIEWER_MISTRAL_API_KEY) and AI_REVIEWER_GITLAB_TOKEN as CI/CD variables in Settings.

PyPI

pip install ai-reviewbot

# Set environment variables (Google)
export AI_REVIEWER_GOOGLE_API_KEY="your-key"
export AI_REVIEWER_GITHUB_TOKEN="your-token"

# Or use Mistral
# export AI_REVIEWER_MISTRAL_API_KEY="your-key"
# export AI_REVIEWER_LLM_PROVIDER=mistral

# Run review
ai-review --repo owner/repo --pr 123

Docker

# DockerHub
docker pull koszivdocker/ai-reviewbot:1

# GitHub Container Registry
docker pull ghcr.io/konstziv/ai-code-reviewer:1

📖 Documentation

Full documentation available in 6 languages:

⚙️ Configuration

Note: At least one LLM API key is required. Old variable names without AI_REVIEWER_ prefix still work as fallback.

See Configuration Guide for all options.

🎯 Example Output

The reviewer provides structured feedback with inline suggestions:

Summary Comment

🤖 AI Code Review

📊 Summary — Found 2 issues and 1 good practice.

Category	Critical	Warning	Info
Security	1	0	0
Code Quality	0	1	0

✨ Good Practices — Excellent error handling in api/handlers.py

---

⏱️ 1.2s | 🪙 1,540 tokens | 💰 ~$0.002

Inline Comment with "Apply" Button

⚠️ SQL Injection Risk

User input is concatenated directly into SQL query.

cursor.execute("SELECT * FROM users WHERE id = ?", (user_id,))

💡 Why this matters: SQL injection allows attackers to execute arbitrary SQL commands. Always use parameterized queries.

📚 Learn more

🛠️ Development

# Clone repository
git clone https://github.com/KonstZiv/ai-code-reviewer.git
cd ai-code-reviewer

# Install dependencies with uv
uv sync --all-groups

# Run tests
uv run pytest

# Run linters
uv run ruff check .
uv run mypy src/

# Build documentation
uv run mkdocs serve

📦 Installation Options

💰 Cost Estimate

Using Gemini 2.5 Flash:

• Input: $0.075 / 1M tokens
• Output: $0.30 / 1M tokens
• Average review: ~$0.002 (1,500 tokens)

100 reviews/month ≈ $0.20

📄 License

Apache 2.0 — See LICENSE for details.

🤝 Contributing

Contributions are welcome! See Contributing Guide.

📬 Support

• 🐛 Report a Bug
• 💡 Request a Feature
• 📚 Documentation

Made with ❤️ by Kostyantin Zivenko