dwisiswant0 / prepare-commit-msg-ai

Prepare Git Commit Message with AI: Write commit message based on code changes with AI.

Prepare Git Commit Message with AI

prepare-commit-msg-ai helps you to write commit message based on code changes with AI.

• Prerequisites
• Installation
  • Usage
• Getting Started
  • Install
  • Uninstall
  • Edit
• License

---

Prerequisites

• go
• git
• mods

Installation

go install github.com/dwisiswant0/prepare-commit-msg-ai@latest

Usage

$ prepare-commit-msg-ai -h

  Prepare Git Commit Message with AI v0.0.1
  --
  made with ♥ by @dwisiswant0

Usage:
  prepare-commit-msg-ai [COMMAND]

Commands:
  install           Install hook in current Git project (alias: isntall, i)
  uninstall         Uninstall generated hook file (alias: unisntall, u)
  edit              Edit hook file (alias: ed, e)

Examples:
  prepare-commit-msg-ai install

Getting Started

Explore the basic functionalities of the tool with the following actions:

Install

To set up this hook, you must individually install it in each Git project, and the installation should take place within the respective project directory. ^{Ain't no way to install this globally unless you make an alias for git commit.}

prepare-commit-msg-ai install

Tip

Alias for install: isntall, i.

Important

Don't forget to set up mods with --settings.

How it works?

$ cd your-git-project/
$ # make changes
$ git add .
$ git commit
$ # generating commit message for ya,
$ # and it'll pop your `EDITOR` to confirm the commit message.
$ # satifying? save it and leave it.

Uninstall

To remove the prepare-commit-msg hook in your current Git project:

prepare-commit-msg-ai uninstall

Tip

Alias for uninstall: unisntall, u.

Edit

Want to change model or prompt?

prepare-commit-msg-ai edit

Tip

Alias for edit: ed, e.

Note

Currently, the hook works correctly when tested with OpenAI's models, using default settings from mods. To clarify, I have NOT tested its compatibility with GPT4All-J * models or models from Azure OpenAI at this time.
^{I mean, I gave the llama2, codellama, openchat, etc. models a shot with different prompts on Ollama, but goddamn the output keeps throwing off the system prompt and always goes off-script, making it a pain for me to create an extractor.}

License

Contributions are welcome! Comments? Feedbacks? Thoughts? Just open an issue.

prepare-commit-msg-ai is made with ♥ by @dwisiswant0 under the MIT license. See LICENSE.