CodingFox: AI Code Reviews

                                        ████                                
                                    ████▒▒██                                
                                  ████  ▒▒██                                
                                ██▒▒  ▒▒▒▒▒▒██                              
                              ██▒▒██        ██                              
  ████                      ██▒▒██          ██                              
██▒▒▒▒██████                ██▒▒██      ▒▒  ████                            
██▒▒▒▒██    ████      ██████▒▒▒▒▒▒██    ▒▒▒▒██████████████                  
██▒▒    ████▒▒▒▒██████▒▒▒▒▒▒▒▒▒▒▒▒▒▒██▒▒▒▒▒▒██▒▒▒▒▒▒▒▒▒▒▒▒████              
██▒▒▒▒      ██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██▒▒██▒▒▒▒▒▒▒▒▒▒▒▒▒▒██            
  ██▒▒      ██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██▒▒██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒████        
  ██        ██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██      
  ██▒▒    ██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒████▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██    
  ██▒▒▒▒  ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒  ██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██    
    ██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒    ██▒▒▒▒▒▒▒▒▒▒████▒▒▒▒▒▒▒▒██  
    ████▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██      ██▒▒▒▒▒▒████▒▒▒▒▒▒▒▒▒▒▒▒██  
    ██▒▒██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██        ██▒▒▒▒██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██  
      ██▒▒██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██        ██████▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██  
      ██▒▒██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██      ██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██
        ████  ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒    ██▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██
          ██    ▒▒██████▒▒▒▒▒▒▒▒▒▒▒▒▒▒    ██▒▒  ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒██
          ██            ████▒▒▒▒▒▒▒▒▒▒    ██  ▒▒  ▒▒        ▒▒▒▒▒▒▒▒▒▒▒▒██  
            ██                      ██  ████  ▒▒          ▒▒▒▒▒▒▒▒▒▒▒▒▒▒██  
              ██                      ██▒▒██              ▒▒  ▒▒▒▒▒▒▒▒▒▒██  
                ██████████████████████▒▒▒▒██                    ▒▒▒▒▒▒██    
                      ██▒▒      ██▒▒▒▒▒▒▒▒██                    ▒▒▒▒██      
                      ██▒▒▒▒  ██▒▒▒▒▒▒▒▒████                  ▒▒▒▒██        
                      ██▒▒▒▒▒▒██▒▒▒▒▒▒██  ██                    ██          
                        ██████▒▒▒▒▒▒██    ██                ████            
                              ██████      ██          ██████                
                                            ██    ████                      
                                            ██████                                   

                        CodingFox - Your AI Code Review Partner
                        Stop shipping bugs. Start shipping excellence.

Overview

CodingFox is an intelligent AI-powered code review assistant that revolutionizes your pull request workflow. Supporting multiple AI providers (OpenAI, Anthropic, and OpenAI-compatible endpoints), CodingFox provides instant, contextual code reviews that catch bugs, improve code quality, and accelerate your development cycle.

Key Features

Multi-Provider Support: OpenAI (GPT-5), Anthropic (Claude), or any OpenAI-compatible API
Instant PR Analysis: Get comprehensive code reviews in seconds
Line-by-Line Review: Detailed suggestions for every code change
Interactive Chat: Tag @codingfox to ask questions about your code
Smart Summaries: Automatic PR summaries and release notes
Customizable: Tailor prompts, models, and review behavior to your needs

Quick Start

1. Add Your API Key to GitHub Secrets

Go to your repository Settings > Secrets and variables > Actions > New repository secret:

Provider	Secret Name	Where to Get It
OpenAI	`OPENAI_API_KEY`	platform.openai.com/api-keys
Anthropic	`ANTHROPIC_API_KEY`	console.anthropic.com
OpenAI-compatible	`AI_API_KEY`	Your provider's dashboard

2. Create the Workflow

Create .github/workflows/codingfox-review.yml:

name: CodingFox AI Review

permissions:
  contents: read
  pull-requests: write

on:
  pull_request:
    types: [opened, synchronize, reopened]
  pull_request_review_comment:
    types: [created]

concurrency:
  group: ${{ github.repository }}-${{ github.event.number || github.head_ref || github.sha }}-${{ github.workflow }}
  cancel-in-progress: ${{ github.event_name != 'pull_request_review_comment' }}

jobs:
  review:
    runs-on: ubuntu-latest
    steps:
      - uses: codingfox/ai-pr-reviewer@latest
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}

3. Create a Pull Request

CodingFox will automatically review your PR within 30-60 seconds!

Provider Configuration

OpenAI (Default)

- uses: codingfox/ai-pr-reviewer@latest
  env:
    GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
    OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
  with:
    ai_provider: openai
    ai_light_model: gpt-5-mini    # Fast, cheap - for summaries
    ai_heavy_model: gpt-5.2       # Powerful - for code review

Anthropic Claude

- uses: codingfox/ai-pr-reviewer@latest
  env:
    GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
    ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
  with:
    ai_provider: anthropic
    ai_light_model: claude-haiku-4-5     # Fast, cheap
    ai_heavy_model: claude-sonnet-4-5    # Powerful

OpenAI-Compatible (OpenRouter, Together, Groq, etc.)

- uses: codingfox/ai-pr-reviewer@latest
  env:
    GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
    AI_API_KEY: ${{ secrets.OPENROUTER_API_KEY }}
  with:
    ai_provider: openai-compatible
    ai_base_url: https://openrouter.ai/api/v1
    ai_light_model: meta-llama/llama-3-8b-instruct
    ai_heavy_model: anthropic/claude-3.5-sonnet

Model Recommendations

OpenAI

Use Case	Model	Cost
Budget-friendly testing	`gpt-5-nano`	$0.05/MTok in
Balanced (default)	`gpt-5-mini` / `gpt-5.2`	$0.25 / $1.75
Maximum quality	`gpt-5.2-pro`	$1.75/MTok in

Anthropic

Use Case	Model	Cost
Fast & cheap	`claude-haiku-4-5`	$1/MTok in
Balanced (recommended)	`claude-sonnet-4-5`	$3/MTok in
Maximum quality	`claude-opus-4-5`	$15/MTok in

Production: Use Pinned Versions

Model aliases (e.g., claude-sonnet-4-5) auto-update when new versions release. For consistent behavior in production, pin to specific versions:

# Development (auto-updates)
ai_heavy_model: claude-sonnet-4-5

# Production (stable)
ai_heavy_model: claude-sonnet-4-5-20250929

OpenAI equivalents: gpt-5.2 vs gpt-5.2-2025-12-11

Configuration Reference

AI Provider Options

Option	Description	Default
`ai_provider`	Provider: `openai`, `anthropic`, or `openai-compatible`	`openai`
`ai_light_model`	Model for summaries and simple tasks	`gpt-5-mini`
`ai_heavy_model`	Model for code review	`gpt-5.2`
`ai_base_url`	Custom API endpoint (for openai-compatible)	-
`ai_temperature`	Response randomness (0.0-1.0)	`0.05`
`ai_timeout_ms`	API timeout in milliseconds	`360000`
`ai_retries`	Number of retry attempts	`5`
`ai_concurrency_limit`	Max concurrent API calls	`6`

Review Behavior

Option	Description	Default
`review_simple_changes`	Review minor changes (typos, formatting)	`false`
`review_comment_lgtm`	Comment when code looks good	`false`
`max_files`	Max files to review (0 = unlimited)	`0`
`path_filters`	Glob patterns for files to include/exclude	-
`debug`	Enable verbose logging	`false`

Custom Prompts

with:
  system_message: |
    You are @codingfox, an expert code reviewer focused on:
    - Security best practices
    - Performance optimization  
    - Clean code principles
    Be constructive and specific in your feedback.

Interactive Chat

Tag @codingfox in any PR comment:

@codingfox Can you suggest test cases for this function?
@codingfox How can I improve the performance here?
@codingfox Is there a security concern with this approach?

Skip Review

Add to your PR description:

@codingfox: ignore

Migration from v0.x

If you're upgrading from an older version using openai_* options, update to the new ai_* options:

Old Option	New Option
`openai_light_model`	`ai_light_model`
`openai_heavy_model`	`ai_heavy_model`
`openai_model_temperature`	`ai_temperature`
`openai_timeout_ms`	`ai_timeout_ms`
`openai_retries`	`ai_retries`
`openai_concurrency_limit`	`ai_concurrency_limit`
`openai_base_url`	`ai_base_url`

The old options still work but will show deprecation warnings.

Troubleshooting

Issue	Solution
CodingFox not commenting	Check Actions tab for errors, verify API key is set
"Rate limit exceeded"	Add credits to your account or reduce `ai_concurrency_limit`
Wrong provider used	Ensure `ai_provider` matches your API key
Timeout errors	Increase `ai_timeout_ms`
Reviews too verbose	Set `review_simple_changes: false`

Development

# Install dependencies
bun install

# Run all checks
bun run all  # typecheck, lint, format, test, build

# Individual commands
bun run typecheck
bun run lint
bun run test
bun run build

License

MIT License - see LICENSE file for details.

CodingFox - Elevating Code Quality, One Review at a Time

RawToast/codingfox

README