Skip to content
iGit Docs

AI Commit Feature

Feature Introduction 🤖

iGit’s AI commit feature uses advanced AI models to analyze your code changes and automatically generate commit messages that comply with the Conventional Commits specification. This not only saves your time but also ensures the quality and consistency of commit messages.

How It Works ⚙️

  1. Collect Changes

    When you run igit commit, iGit collects all changes in the Git staged area.

  2. Analyze Changes

    The AI model analyzes these changes, including:

    • Modified file types
    • Code additions, deletions, and modifications
    • Change context

  3. Generate Commit Message

    Based on the analysis, AI generates a commit message following the conventional commits specification, including:

    • Type (feat/fix/docs etc.)
    • Scope (optional)
    • Description
    • Detailed explanation (if needed)

  4. Confirm Commit

    You can choose to accept or modify the generated commit message.

Usage 📝

Basic Usage

Terminal window
# Generate commit message and wait for confirmation
igit commit


# Generate commit message and commit directly
igit commit -y


# Generate message only without git commit
igit commit -d
# or
igit commit --dry-run

Commit Types

AI will automatically select the most appropriate commit type based on the changes:

  • feat: New feature
  • fix: Bug fix
  • docs: Documentation updates
  • style: Code style changes
  • refactor: Code refactoring
  • perf: Performance improvements
  • test: Test-related changes
  • chore: Build process or auxiliary tool changes

Best Practices 💡

1. Keep Changes Focused

Include only related changes in each commit, so AI can better understand your intentions.

2. Use Staged Area

Make sure all files you want to commit are added to the staged area:

Terminal window
git add <files>
igit commit

3. Review Generated Messages

Although AI usually generates high-quality commit messages, it’s best to quickly review them to ensure they accurately reflect your intentions.

Common Questions ❓

Q: Why are generated commit messages sometimes not accurate enough?

A: This might be due to:

  • Changes being too complex or scattered
  • Staged area containing unrelated changes
  • AI model unable to fully understand specific business context

Q: How to improve the quality of generated commit messages?

  1. Ensure each commit contains only related changes