Code-Tanuki - AI-Powered Code Review

Intelligent PR review automation powered by Claude AI & Google Gemini, seamlessly integrated with GitHub and Jira.

Overview

Code-Tanuki is an AI-powered code review tool designed for development teams. It leverages Claude AI or Google Gemini to provide intelligent, context-aware code reviews on your GitHub pull requests, with optional Jira integration for requirement validation.

Key Features

🤖 Multi-AI Support - Choose between Claude AI (Opus, Sonnet, Haiku) or Google Gemini (3 Pro, 3 Flash)
🔗 GitHub Integration - Automatic PR fetching and inline comment posting
📋 Jira Integration - Validate code against acceptance criteria (optional)
✨ Beautiful UI - Modern, colorful interface with smooth animations
📊 Review History - Track all reviews with filtering and search capabilities
🐳 Docker Ready - Easy deployment with Docker Compose
🔄 Smart Retry Logic - Automatic retry with exponential backoff
📝 Production Logging - Winston-based structured logging
🎯 Type Safe - Full TypeScript with Zod validation

Tech Stack

Frontend

Next.js 16 with App Router
React 19 with TypeScript
Material-UI (MUI) for component library
Tailwind CSS for utility styling
Emotion for styled components

Backend

Next.js API Routes for serverless functions
Anthropic SDK for Claude AI integration
Google Generative AI SDK for Gemini integration
Octokit for GitHub API
Axios for Jira REST API
Winston for logging
Zod for runtime validation

Infrastructure

Docker for containerization
Docker Compose for orchestration
JSON Storage with abstraction for future DB migration

Prerequisites

Before you begin, you'll need:

Node.js 20+ and npm
AI Provider API Key (at least one required):
- Anthropic API Key for Claude models
  - Get from: https://console.anthropic.com/settings/keys
  - See Setup Guide
- Google Gemini API Key for Gemini models (optional alternative)
  - Get from: https://aistudio.google.com/app/apikey
  - Free tier available
GitHub Personal Access Token (required)
- Fine-grained token with "Pull requests: Read and write"
- See Setup Guide
Jira API Token (optional)
- See Setup Guide

Quick Start

1. Clone and Install

git clone https://github.com/hongjs/code-tanuki.git
cd code-tanuki
npm install

2. Configure Environment

cp .env.example .env

Edit .env and add your credentials:

# AI Provider - At least one required
ANTHROPIC_API_KEY=sk-ant-your_api_key_here  # For Claude models
GEMINI_API_KEY=your_gemini_api_key_here      # For Gemini models (optional)

# Required - GitHub
GITHUB_TOKEN=ghp_your_github_token_here

# Optional - Jira (leave blank to skip Jira integration)
JIRA_BASE_URL=https://yourcompany.atlassian.net
JIRA_EMAIL=your-email@company.com
JIRA_API_TOKEN=your_jira_token_here

3. Create Required Directories

mkdir -p data/reviews logs
echo "[]" > data/reviews/all-reviews.json

4. Run Development Server

npm run dev

Open http://localhost:3000 and you're ready to go!

Docker Deployment

Using Docker Compose (Recommended)

# Build and start
docker-compose up -d

# View logs
docker-compose logs -f

# Stop
docker-compose down

Manual Docker Build

# Build image
docker build -t code-tanuki:latest .

# Run container
docker run -p 3000:3000 \
  --env-file .env \
  -v ./data:/app/data \
  -v ./logs:/app/logs \
  code-tanuki:latest

Usage

Starting a Code Review

Navigate to the Review page
Enter a GitHub PR URL:
```
https://github.com/owner/repo/pull/123
```
(Optional) Enter a Jira ticket ID or leave blank for auto-extraction
(Optional) Add custom instructions for the AI reviewer
Select your preferred AI model (Claude or Gemini)
Click Start Review

Review Process

Code-Tanuki will:

Fetch PR Data - Download diff, files, and metadata from GitHub
Extract Jira Context - Get ticket details and acceptance criteria (if configured)
AI Analysis - Your chosen AI model (Claude or Gemini) analyzes the code for:
- Bugs and potential issues
- Security vulnerabilities
- Code quality and best practices
- Requirement compliance
Preview - Review AI comments before posting
Post Comments - Inline comments added to GitHub PR
Update Jira - Post review summary to Jira ticket (if configured)

Jira Ticket Auto-Extraction

Code-Tanuki automatically extracts Jira ticket IDs from PR titles:

feat(BYD-1234): Add user authentication → Extracts BYD-1234
fix(PROJ-567): Fix login bug → Extracts PROJ-567

Supported prefixes: feat, fix, chore, docs, style, refactor, test, build, ci, perf

Viewing History

Navigate to the History page to:

View all completed reviews
Filter by status, model, or date
Search by PR number or repository
Click any review for full details

Project Structure

code-tanuki/
├── src/
│   ├── app/                      # Next.js App Router
│   │   ├── review/              # Review page
│   │   ├── history/             # History page
│   │   ├── layout.tsx           # Root layout with sidebar
│   │   └── api/                 # API Routes
│   │       ├── review/          # Main orchestrator
│   │       ├── github/          # GitHub endpoints
│   │       ├── jira/            # Jira endpoints
│   │       └── health/          # Health check
│   ├── components/              # React Components
│   │   ├── review/             # Review form & progress
│   │   ├── history/            # History table & filters
│   │   └── layout/             # Sidebar navigation
│   ├── lib/                    # Business Logic
│   │   ├── api/                # API clients (GitHub, Jira, Claude)
│   │   ├── storage/            # Storage abstraction
│   │   ├── utils/              # Utilities & helpers
│   │   ├── logger/             # Winston logger
│   │   └── constants/          # Constants, prompts, models
│   └── types/                  # TypeScript Types
├── data/                       # Review data (JSON files)
├── logs/                       # Application logs
├── docs/                       # Documentation
├── Dockerfile                  # Docker configuration
├── docker-compose.yml          # Docker Compose setup
└── .env                        # Environment variables

Configuration

Environment Variables

Required

GITHUB_TOKEN - GitHub Personal Access Token

AI Provider (At least one required)

Claude AI:

ANTHROPIC_API_KEY - Your Claude API key (starts with sk-ant-)
CLAUDE_MODEL_DEFAULT - Default Claude model (default: claude-haiku-4-5-20251001)
CLAUDE_MAX_TOKENS - Max output tokens (default: 8192)
CLAUDE_TEMPERATURE - AI temperature 0-1 (default: 0.3)

Google Gemini (Optional):

GEMINI_API_KEY - Your Gemini API key (optional)
GEMINI_MODEL_DEFAULT - Default Gemini model (default: gemini-2.0-flash)
GEMINI_MAX_TOKENS - Max output tokens (default: 2048, recommended for free tier)
GEMINI_TEMPERATURE - AI temperature 0-2 (default: 0.3)

Note for Free Tier Users: The free tier works best with GEMINI_MAX_TOKENS=2048 or lower to avoid response truncation. For larger PRs, consider using Claude or upgrading to Gemini paid tier.

Optional (Jira)

JIRA_BASE_URL - Your Jira instance URL
JIRA_EMAIL - Your Jira account email
JIRA_API_TOKEN - Jira API token

Storage & Logging

DATA_DIR - Review storage directory (default: ./data/reviews)
LOG_DIR - Log directory (default: ./logs)
LOG_LEVEL - Logging level: debug | info | warn | error (default: info)

Retry Configuration

RETRY_MAX_ATTEMPTS - Max retry attempts (default: 3)
RETRY_BASE_DELAY_MS - Base delay between retries (default: 1000)
RETRY_MAX_DELAY_MS - Max delay between retries (default: 10000)

Available AI Models

Claude Models (Anthropic)

Model	ID	Best For	Speed	Max Tokens
Claude Opus 4.5	`claude-opus-4-5-20251101`	Complex reviews, highest accuracy	Slower	8192
Claude Sonnet 4.5	`claude-sonnet-4-5-20250929`	Balanced performance	Medium	8192
Claude Haiku 4.5	`claude-haiku-4-5-20251001`	Quick reviews, simple PRs	Fastest	8192

Gemini Models (Google)

Model	ID	Best For	Speed	Max Tokens*
Gemini 3 Pro Preview	`gemini-3-pro-preview`	Complex reasoning, thorough reviews	Medium	2048**
Gemini 3 Flash Preview	`gemini-3-flash-preview`	Fast reviews, efficient	Fast	2048**

Note: You can use either Claude or Gemini models. Configure the appropriate API key in your .env file.

* Max output tokens (configurable) ** Recommended setting for free tier to avoid response truncation. Paid tier supports higher limits.

Getting Your Gemini API Key (Free Tier Available!)

Google Gemini offers a free tier that's perfect for trying out Code-Tanuki:

Visit Google AI Studio: https://aistudio.google.com/app/apikey
Sign in with your Google account
Click "Create API Key"
Copy your API key and add it to .env:
```
GEMINI_API_KEY=your_gemini_api_key_here
```

Free Tier Limits:

15 requests per minute
1 million tokens per minute
1,500 requests per day
Recommended max output tokens: 2048

Important: To avoid response truncation on the free tier, make sure to set:

GEMINI_MAX_TOKENS=2048

For larger PRs (>500 lines), you may get partial reviews. Consider:

Using Claude AI instead
Breaking PRs into smaller chunks
Upgrading to Gemini's paid tier

Choosing Between Claude and Gemini

Feature	Claude	Gemini
Cost	Paid (usage-based)	Free tier available
Best Models	Opus 4.5, Sonnet 4.5	Gemini 3 Pro Preview
Speed	Very fast (Haiku)	Very fast (Flash)
Code Understanding	Excellent	Excellent
Context Window	200K tokens	2M tokens (Gemini 3)
Best For	Production workloads	Testing, small teams

Recommendation:

Start with Gemini (free tier) to test Code-Tanuki
Upgrade to Claude for production use or larger teams

Storage

Code-Tanuki uses JSON file storage by default with an abstraction layer for easy database migration:

Individual Reviews: data/reviews/{timestamp}-{prNumber}.json
Aggregated Index: data/reviews/all-reviews.json

Migrating to a Database

Implement IStorageAdapter interface for your database
Update STORAGE_TYPE environment variable
No application code changes needed!

Logging

Logs are written to:

logs/combined.log - All logs
logs/error.log - Error logs only
Console (development mode only)

Log levels: debug | info | warn | error

Available Scripts

Development

npm run dev          # Start dev server with hot reload
npm run build        # Build for production
npm run start        # Start production server
npm run lint         # Run ESLint
npm run type-check   # TypeScript type checking
npm run format       # Format code with Prettier
npm run format:check # Check code formatting

Docker

npm run docker:build         # Build Docker image
npm run docker:run           # Run Docker container
npm run docker:compose:up    # Start with Docker Compose
npm run docker:compose:down  # Stop Docker Compose
npm run docker:compose:logs  # View Docker logs

API Endpoints

Health Check

GET /api/health

Review

POST /api/review
Content-Type: application/json

{
  "prUrl": "https://github.com/owner/repo/pull/123",
  "jiraTicketId": "BYD-1234",  // optional
  "additionalPrompt": "...",    // optional
  "modelId": "claude-opus-4-20250514",
  "previewOnly": true
}

Submit Review

POST /api/review/submit
Content-Type: application/json

{
  "prUrl": "https://github.com/owner/repo/pull/123",
  "jiraTicketId": "BYD-1234",  // optional
  "modelId": "claude-opus-4-20250514",
  "comments": [...]
}

History

GET /api/history

Troubleshooting

Build Errors

rm -rf node_modules package-lock.json .next
npm install
npm run build

Environment Variable Issues

Make sure .env exists and contains all required variables:

cp .env.example .env
# Edit .env and fill in your API keys

Gemini Response Truncation (Free Tier)

Error: Failed to parse Gemini response as JSON. Response may be truncated.

Solution: The free tier has token limits. Update your .env:

# Reduce max tokens for free tier
GEMINI_MAX_TOKENS=2048

# Or use an even lower value for very large PRs
GEMINI_MAX_TOKENS=1024

Alternative: Switch to Claude AI for larger PRs:

ANTHROPIC_API_KEY=sk-ant-your_key_here

Then select a Claude model in the UI instead of Gemini.

Storage Errors

Ensure directories exist:

mkdir -p data/reviews logs
echo "[]" > data/reviews/all-reviews.json

Docker Issues

# Clean and rebuild
docker-compose down -v
docker-compose build --no-cache
docker-compose up -d

Documentation

Contributing

Contributions are welcome! Please:

Fork the repository
Create a feature branch
Make your changes
Submit a pull request

License

ISC

Support

For issues and questions:

GitHub Issues: https://github.com/hongjs/code-tanuki/issues
Documentation: See /docs folder

Made with ❤️ by the Code-Tanuki team

Name		Name	Last commit message	Last commit date
Latest commit History 17 Commits
docs		docs
src		src
.aiignore		.aiignore
.dockerignore		.dockerignore
.env.example		.env.example
.eslintrc.json		.eslintrc.json
.gitignore		.gitignore
.prettierrc		.prettierrc
CLAUDE.md		CLAUDE.md
Dockerfile		Dockerfile
LICENSE		LICENSE
README.md		README.md
docker-compose.yml		docker-compose.yml
next-env.d.ts		next-env.d.ts
next.config.js		next.config.js
package.json		package.json
postcss.config.js		postcss.config.js
tailwind.config.js		tailwind.config.js
tsconfig.json		tsconfig.json
yarn.lock		yarn.lock

License

hongjs/code-tanuki

Folders and files

Latest commit

History

Repository files navigation

Code-Tanuki - AI-Powered Code Review

Overview

Key Features

Tech Stack

Frontend

Backend

Infrastructure

Prerequisites

Quick Start

1. Clone and Install

2. Configure Environment

3. Create Required Directories

4. Run Development Server

Docker Deployment

Using Docker Compose (Recommended)

Manual Docker Build

Usage

Starting a Code Review

Review Process

Jira Ticket Auto-Extraction

Viewing History

Project Structure

Configuration

Environment Variables

Required

AI Provider (At least one required)

Optional (Jira)

Storage & Logging

Retry Configuration

Available AI Models

Claude Models (Anthropic)

Gemini Models (Google)

Getting Your Gemini API Key (Free Tier Available!)

Choosing Between Claude and Gemini

Storage

Migrating to a Database

Logging

Available Scripts

Development

Docker

API Endpoints

Health Check

Review

Submit Review

History

Troubleshooting

Build Errors

Environment Variable Issues

Gemini Response Truncation (Free Tier)

Storage Errors

Docker Issues

Documentation

Contributing

License

Support

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Contributors 2

Uh oh!

Languages

Packages