EverGit

An AI-powered Git helper for the Evergreen ILS project. Evergit uses OpenAI or Ollama models to generate commit messages that adhere to specific standards, streamlining the commit process while reducing manual input.

Features

• Generate commit messages using OpenAI's LLM models or local Ollama models.
• Support for multiple AI providers with configurable defaults.
• Automatically reference Launchpad bugs in commit messages.
• Automatically sign off commits with the user's name and email.
• Select files to stage for commit.
• Use different models by specifying the model name as an argument.
• Manage user-specific configuration for name, email, and AI provider settings.

Requirements

• Node.js, npm, and Git must be installed on your system.
• For OpenAI: OpenAI API key (stored in the OPENAI_API_KEY environment variable)
• For Ollama: Ollama running locally or accessible via network (optional: OLLAMA_API_KEY for authenticated instances)
• A launchpad account is required to reference bugs in commit messages.

Installation

Install evergit globally using npm:

npm install -g evergit

Usage

evergit needs to be run in a git repository. It will automatically detect the repository and branch you are on. If run without any arguments, it will display the help message.

Commands

evergit commit

Generates a commit message using a LLM model that follows the Evergreen ILS commit message format.

evergit commit  # Uses the default provider/model and prompts the user to select files to stage

• Prompts the user to select files to stage.
• Prompts the user for a Launchpad bug number.
  • References the bug name, description and conversation when generating the commit message using the Launchpad API.
• Automatically signs off the commit with the user's name and email from the git configuration.

Options

• -m, --model <model-name>: Use a specific model to generate the commit message.

  evergit commit --model <model-name> # Uses a specific model

• -a, --all: Stage all modified files for commit.

  evergit commit --all    # Stages all modified files

• -p, --provider <provider>: Use a specific AI provider (openai or ollama).

  evergit commit --provider ollama    # Uses Ollama instead of default provider

evergit config

Manage user-specific configuration options for name, email, and AI provider settings. If no name and/or email is set, evergit defaults to using the git configuration.

Options

• --setup-provider: Interactive setup for AI provider configuration.

  evergit config --setup-provider

• --set <key>: Set a configuration option. Valid keys are name, email, provider, openaiModel, ollamaModel, and ollamaBaseUrl.

  evergit config --set name "Your Name"
  evergit config --set email "your.email@example.com"
  evergit config --set provider "ollama"
  evergit config --set ollamaModel "llama3.2"
  evergit config --set ollamaBaseUrl "http://localhost:11434"

• --get <key>: Get a configuration option.

  evergit config --get provider
  evergit config --get ollamaModel

• --clear <key>: Clear a configuration option.

  evergit config --clear provider

• --get-all: Get the entire configuration.

  evergit config --get-all

• --edit: Edit the configuration file manually in the default editor.

  evergit config --edit

AI Provider Configuration

OpenAI (Default)

Evergit uses OpenAI by default. Ensure you have the OPENAI_API_KEY environment variable set.

Ollama

To use Ollama models:

1. Install and run Ollama locally (see ollama.ai)
2. Configure evergit to use Ollama:

   evergit config --setup-provider
   # Select "ollama" and follow the prompts

3. Or set manually:

   evergit config --set provider ollama
   evergit config --set ollamaBaseUrl "http://localhost:11434"
   evergit config --set ollamaModel "llama3.2"