Installation¶

The installation option depends on your use case and goals.

1. CI/CD — Automated Review¶

The most common scenario: AI Code Reviewer runs automatically when a PR/MR is created or updated.

Set up in 5 minutes:

2. Standalone Deployment: CLI/Docker¶

CLI and Docker image allow running AI Code Reviewer outside the standard CI pipeline.

Use Cases¶

Scenario	How to implement
Manual run	Local terminal — debugging, demo, evaluation
Scheduled review	GitLab Scheduled Pipeline / GitHub Actions `schedule` / cron
Batch review	Script iterating over open PR/MR
Own server	Docker on server with Git API access
On-demand review	Webhook → container launch

Required Environment Variables¶

Variable	Description	When needed	How to get
`AI_REVIEWER_GOOGLE_API_KEY`	Gemini API key	Always	Google AI Studio
`AI_REVIEWER_GITHUB_TOKEN`	GitHub Personal Access Token	For GitHub	Instructions
`AI_REVIEWER_GITLAB_TOKEN`	GitLab Personal Access Token	For GitLab	Instructions

Fallback

Old names without prefix (e.g., GOOGLE_API_KEY) still work as fallback.

Manual Run¶

For debugging, demo, evaluation before deployment, retrospective PR/MR analysis.

Docker (recommended)¶

No Python installation required — everything is in the container.

Step 1: Pull the image

docker pull ghcr.io/konstziv/ai-code-reviewer:1

Step 2: Run the review

GitHub PRGitLab MR

docker run --rm \
  -e AI_REVIEWER_GOOGLE_API_KEY=your_api_key \
  -e AI_REVIEWER_GITHUB_TOKEN=your_token \
  ghcr.io/konstziv/ai-code-reviewer:1 \
  --repo owner/repo --pr 123

docker run --rm \
  -e AI_REVIEWER_GOOGLE_API_KEY=your_api_key \
  -e AI_REVIEWER_GITLAB_TOKEN=your_token \
  ghcr.io/konstziv/ai-code-reviewer:1 \
  --provider gitlab --repo owner/repo --pr 123

Docker images

Available from two registries:

ghcr.io/konstziv/ai-code-reviewer:1 — GitHub Container Registry
koszivdocker/ai-reviewbot:1 — DockerHub

pip / uv¶

Installation as a Python package.

Step 1: Install

pipuvpipx

pip install ai-reviewbot

uv tool install ai-reviewbot

pipx install ai-reviewbot

Python version

Requires Python 3.13+

Step 2: Set up variables

export AI_REVIEWER_GOOGLE_API_KEY=your_api_key
export AI_REVIEWER_GITHUB_TOKEN=your_token  # or AI_REVIEWER_GITLAB_TOKEN for GitLab

Step 3: Run

GitHub PRGitLab MR

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

ai-review --provider gitlab --repo owner/repo --pr 123

Optional Variables¶

Additional variables are available for fine-tuning:

Variable	Default	Effect
`AI_REVIEWER_LANGUAGE`	`en`	Response language (ISO 639)
`AI_REVIEWER_LANGUAGE_MODE`	`adaptive`	Language detection mode
`AI_REVIEWER_GEMINI_MODEL`	`gemini-2.5-flash`	Gemini model
`AI_REVIEWER_LOG_LEVEL`	`INFO`	Logging level

Full list of variables →

Scheduled Reviews¶

Running reviews on a schedule — for resource savings or when instant feedback is not needed.

GitLab Scheduled PipelineGitHub Actions Schedule

# .gitlab-ci.yml
ai-review-scheduled:
  image: ghcr.io/konstziv/ai-code-reviewer:1
  script:
    - |
      # Get list of open MRs
      MR_LIST=$(curl -s --header "PRIVATE-TOKEN: $AI_REVIEWER_GITLAB_TOKEN" \
        "$CI_SERVER_URL/api/v4/projects/$CI_PROJECT_ID/merge_requests?state=opened" \
        | jq -r '.[].iid')

      # Run review for each MR
      for MR_IID in $MR_LIST; do
        echo "Reviewing MR !$MR_IID"
        ai-review --provider gitlab --repo $CI_PROJECT_PATH --pr $MR_IID || true
      done
  rules:
    - if: $CI_PIPELINE_SOURCE == "schedule"

Schedule setup: Project → Build → Pipeline schedules → New schedule

# .github/workflows/scheduled-review.yml
name: Scheduled AI Review

on:
  schedule:
    - cron: '0 9 * * *'  # Daily at 9:00 UTC

jobs:
  review-open-prs:
    runs-on: ubuntu-latest
    steps:
      - name: Get open PRs and review
        env:
          AI_REVIEWER_GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
          AI_REVIEWER_GOOGLE_API_KEY: ${{ secrets.AI_REVIEWER_GOOGLE_API_KEY }}
        run: |
          # Get list of open PRs
          PRS=$(gh pr list --repo ${{ github.repository }} --state open --json number -q '.[].number')

          for PR in $PRS; do
            echo "Reviewing PR #$PR"
            docker run --rm \
              -e AI_REVIEWER_GOOGLE_API_KEY -e AI_REVIEWER_GITHUB_TOKEN \
              ghcr.io/konstziv/ai-code-reviewer:1 \
              --repo ${{ github.repository }} --pr $PR || true
          done

Own Server / Private Environment¶

For deployment on your own infrastructure with Git API access.

Options:

Docker on server — run via cron, systemd timer, or as a service
Kubernetes — CronJob for scheduled reviews
Self-hosted GitLab — add GITLAB_URL variable (see example below)

Cron job example:

# /etc/cron.d/ai-review
# Daily at 10:00 run review for all open MRs
0 10 * * * reviewer /usr/local/bin/review-all-mrs.sh

#!/bin/bash
# /usr/local/bin/review-all-mrs.sh
export AI_REVIEWER_GOOGLE_API_KEY="your_key"
export AI_REVIEWER_GITLAB_TOKEN="your_token"

MR_LIST=$(curl -s --header "PRIVATE-TOKEN: $AI_REVIEWER_GITLAB_TOKEN" \
  "https://gitlab.company.com/api/v4/projects/123/merge_requests?state=opened" \
  | jq -r '.[].iid')

for MR_IID in $MR_LIST; do
  docker run --rm \
    -e AI_REVIEWER_GOOGLE_API_KEY -e AI_REVIEWER_GITLAB_TOKEN \
    ghcr.io/konstziv/ai-code-reviewer:1 \
    --provider gitlab --repo group/repo --pr $MR_IID
done

Self-hosted GitLab

For self-hosted GitLab add the AI_REVIEWER_GITLAB_URL variable:

-e AI_REVIEWER_GITLAB_URL=https://gitlab.company.com

3. Contributors / Development¶

If you have the time and inspiration to help develop the package, or want to use it as a foundation for your own development — we sincerely welcome and encourage such actions!

Development Installation¶

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

# Install dependencies (we use uv)
uv sync

# Verify
uv run ai-review --help

# Run tests
uv run pytest

# Run quality checks
uv run ruff check .
uv run mypy .

We use uv for dependency management.

Install: curl -LsSf https://astral.sh/uv/install.sh | sh

Project Structure¶

ai-code-reviewer/
├── src/ai_reviewer/      # Source code
│   ├── core/             # Models, config, formatting
│   ├── integrations/     # GitHub, GitLab, Gemini
│   └── utils/            # Utilities
├── tests/                # Tests
├── docs/                 # Documentation
└── examples/             # CI configuration examples

How to contribute →

Requirements¶

System Requirements¶

Component	Requirement
Python	3.13+ (for pip install)
Docker	20.10+ (for Docker)
OS	Linux, macOS, Windows
RAM	256MB+
Network	Access to `generativelanguage.googleapis.com`

API Keys¶

Key	Required	How to get
Google Gemini API	Yes	Google AI Studio
GitHub PAT	For GitHub	Instructions
GitLab PAT	For GitLab	Instructions

Gemini API Limits¶

Free tier

Google Gemini has a free tier:

Limit	Value
Requests per minute	15 RPM
Tokens per day	1M
Requests per day	1500

This is sufficient for most projects.

Next Step¶

Quick Start →