AI Git

AI-powered Git commit messages using a local LLM

Usage

Requirements

• llama.cpp running locally (e.g. ./llama-server --port 8080)

Install

gem install ai_git

Environment Variables

Variable	Description	Default
AI_GIT_MODEL_NAME	Model name	ggml-org/gemma-4-E4B-it-GGUF:Q8_0
AI_GIT_BASE_URL	Base URL of the llama.cpp server	http://127.0.0.1:8080
NO_COLOR	Disable colored terminal output when set	—

Run ai_git config to see exactly which model, URL and endpoint are resolved from your environment.

ai_git talks to a local llama.cpp server over its OpenAI-compatible /v1/chat/completions endpoint. No API key is needed.

Example

# Start llama.cpp's server, then run ai_git with defaults
./llama-server --port 8080
ai_git

# Or point at a custom model/port
export AI_GIT_MODEL_NAME=my-model
export AI_GIT_BASE_URL=http://127.0.0.1:8081

Run

git add <files>
ai_git

ai_git generates a commit message from your staged changes, commits, and pushes to the current branch's upstream — no prompts, no flags.

Subcommands

Subcommand	Description
ai_git	Generate a commit message, commit, and push staged files
ai_git config	Show the resolved provider configuration
ai_git --help	Show usage
ai_git --version	Print version

Development

gem build ai_git.gemspec
gem install ./ai_git-$(ruby -r./lib/ai_git/version -e 'print AIGit::VERSION').gem