aicommit

🤖 AI-powered git commit message generator with multi-provider support.

Generate conventional commit messages using AI from OpenAI, Anthropic Claude, Ollama, vLLM, or Kimi.

Features

• Multi-provider: OpenAI, Anthropic Claude, Ollama, vLLM, Kimi
• Conventional Commits: Follows the Conventional Commits specification
• Persistent config: Save your preferred provider and settings
• Interactive mode: Preview, edit, or cancel before committing
• Dry run: Test without creating commits

Installation

npm install -g aicommit-cli

Or link it:

cd /path/to/aicommit
npm link

Quick Start

# 1. Set your default provider
aicommit config set provider vllm

# 2. Add files to staging
git add .

# 3. Generate commit message
aicommit

Usage

Basic Commands

aicommit                  # Generate commit with default provider
aicommit --dry-run        # Preview without committing
aicommit -p openai        # Use specific provider
aicommit -m gpt-4o        # Use specific model

Configuration

# Set default provider
aicommit config set provider vllm

# Set default model
aicommit config set model gpt-4o

# Set temperature (creativity: 0-1)
aicommit config set temperature 0.3

# View current config
aicommit config get

# View specific setting
aicommit config get provider

# Delete a setting
aicommit config delete model

Options

| Option | Description | Default | |--------|-------------|---------| | -p, --provider <name> | AI provider | config/provider or 'openai' | | -m, --model <name> | Model name | config/model | | -t, --temperature <number> | AI creativity (0-1) | 0.3 | | --dry-run | Preview only, no commit | false | | --no-edit | Skip edit confirmation | false |

Providers

OpenAI

export OPENAI_API_KEY=sk-...
aicommit -p openai -m gpt-4o

Models: gpt-4o, gpt-4-turbo, gpt-3.5-turbo

Anthropic Claude

export ANTHROPIC_API_KEY=sk-ant-...
aicommit -p anthropic -m claude-sonnet-4-20250514

Models: claude-sonnet-4-20250514, claude-3-5-sonnet-20241022, claude-3-opus-20240229

Ollama (Local)

export OLLAMA_HOST=http://localhost:11434
aicommit -p ollama -m llama3

Runs open-source models locally. Default port: 11434

Models: llama3, llama3.1, mistral, codellama, etc.

vLLM (Local)

export VLLM_HOST=http://localhost:8090
aicommit -p vllm

OpenAI-compatible local inference server. Default port: 8090

# Example: Run vLLM with Docker
docker run --gpus all -v ~/.cache/huggingface:/root/.cache/huggingface \
  -p 8090:8000 \
  --env "HUGGING_FACE_HUB_TOKEN=hf_..." \
  vllm/vllm-openai:latest \
  --model meta-llama/Llama-3.3-70B-Instruct

Kimi (Moonshot)

export MOONSHOT_API_KEY=sk-...
aicommit -p kimi -m kimi-k2.5

Models: kimi-k2.5, kimi-k2, kimi-k2-thinking

Environment Variables

| Variable | Description | Provider | |----------|-------------|----------| | OPENAI_API_KEY | OpenAI API key | openai | | ANTHROPIC_API_KEY | Anthropic API key | anthropic | | MOONSHOT_API_KEY | Moonshot API key | kimi | | OLLAMA_HOST | Ollama server URL | ollama | | VLLM_HOST | vLLM server URL | vllm |

Configuration File

Settings are stored in ~/.aicommit.json:

{
  "provider": "vllm",
  "model": "llama3",
  "temperature": 0.3
}

Commit Format

Messages follow Conventional Commits:

feat(api): add user authentication endpoint

Implemented JWT-based authentication with refresh tokens.
Includes login, logout, and token refresh endpoints.

Examples

# Quick commit with defaults
git add . && aicommit

# Preview only
aicommit --dry-run

# Use different provider just once
aicommit -p anthropic

# Use specific model
aicommit -m gpt-4-turbo

# More creative responses
aicommit -t 0.7

# Deterministic responses
aicommit -t 0.1

# Skip confirmation prompt
aicommit --no-edit

Requirements

• Node.js >= 18
• Git repository with staged changes
• API key (for cloud providers) or local AI server (Ollama/vLLM)

License

MIT