AI Reviewer (air)

A high-performance CLI tool for automated code reviews. It strips comments from your source code to optimize token usage and privacy, then generates detailed Markdown reports using LLMs.

Installation

Install ai_reviewer and the air shortcut with a single command :

curl -sSL https://raw.githubusercontent.com/realgetOff/AI_Reviewer/main/install.sh | bash

Note: Restart your terminal or run source ~/.zshrc after installation.

Configuration

Set your API key and preferences directly in your terminal :

air config

Commands

Command	Description
air	Analyze current directory
air <file>	Analyze a specific file or path
air config	Edit global configuration (~/.ai_config.json)
air -clean	Run the cleanup script (remove reports dir.)
air -update	Update to the latest version from GitHub
air -delete	Fully uninstall the program and its config

Options

Flag	Description
-s <style>	Set review style (e.g., minimal, security, audit, crash, fix)
-l <lang>	Set output language (en, fr)
-f <fmt>	Set default format (md, pdf)
-t <sec>	Set custom request timeout (default: 30s)
-i <n>	Set max agent iterations (default: 5)
-I <sec>	Set interactive program timeout (default: 10s)
-c <prompt>	Use custom prompt instead of config style
-m	List available AI models
-d	Enable debug mode (verbose logs)
-g	Analyze all files as one (global context mode)
-a	Enable agent mode (AI executes commands autonomously)
--version	Show current version
-h	Display help menu

Output

Output is saved in the reports/ directory:

reports/*.md
reports/*.pdf
reports/error.md
reports/debug.log

AI Providers

Use this guide to configure your config.json. Copy the API URL and the Model Type corresponding to your preferred AI service.

Provider	Model Type	API URL	Key Format
Google Gemini	gemini	https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash:generateContent	URL Param
OpenAI (GPT-4o)	openai	https://api.openai.com/v1/chat/completions	Bearer Token
Anthropic (Claude 3.5)	claude	https://api.anthropic.com/v1/messages	x-api-key
Mistral AI	mistral	https://api.mistral.ai/v1/chat/completions	Bearer Token
Zhipu AI	glm	https://open.bigmodel.cn/api/paas/v4/chat/completions	Bearer Token

Note: For Gemini only, you must leave model_name: "" (empty) because the model is already specified within the API URL (ofc you can change gemini-2.5-flash by any of available models (check air -m))

Configuration Example (Gemini)

If you want to use Gemini-2.5-Flash, your config.json should look like this:

{
  "api_key": "copy-your-gemini-api-key-here",
  "api_url": "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash:generateContent",
  "model_type": "gemini",
  "model_name": "",
  "default_style": "minimal",
  "language": "en",
  "styles":
  {
    "minimal":
    {
      "en": "Identify critical logic bugs.",
      "fr": "Identifie les bugs logiques critiques."
    }
  }
}

(use the chinese model, alternative to claude and almost free;)