git-roast

A command-line tool that uses Google's Gemini API or local LM Studio model to roast your code diffs.

Installation

go install github.com/c0ze/git-roast@latest

Configuration

You can configure the tool using environment variables or a .env file in the current directory.

Variable	Description	Default
ROAST_MODE	GEMINI or LOCAL	GEMINI
GEMINI_API_KEY	Your Google Gemini API Key	Required for Gemini
ROAST_LOCAL_URL	URL for local model (e.g., LM Studio)	http://localhost:1234/v1
ROAST_LOCAL_MODEL	Model name for local inference	deepseek-r1-distill-qwen-7b
SHOW_THINKING	Show "Chain of Thought" for reasoning models	false

Usage

1. Gemini Mode (Default)

1. Get a Gemini API Key: Get an API key from Google AI Studio.
2. Set the Environment Variable:

   export GEMINI_API_KEY="your_api_key_here"

3. Roast Your Code:

   git diff | git-roast

2. Local Mode (e.g., LM Studio)

1. Start your Local Server (e.g., LM Studio on port 1234).
2. Configure .env:

   ROAST_MODE=LOCAL
   ROAST_LOCAL_URL=http://localhost:1234/v1
   ROAST_LOCAL_MODEL=deepseek-r1-distill-qwen-7b
   SHOW_THINKING=false # Set to true to see the reasoning process

3. Roast Your Code:

   git diff | git-roast

How it Works

It reads the diff from stdin, sends it to the configured LLM (Gemini or Local) with a "cynical senior engineer" persona, and prints the critique to stdout.