Fuckmit - AI-Powered Git Commit Message Generator

Read this in other languages: 简体中文

A command-line tool written in Rust that automatically analyzes code changes and generates standardized Git commit messages by integrating with various AI providers (OpenAI, Azure OpenAI, DeepSeek, Qwen, etc.).

Installation

Binary Releases

For Windows, Mac OS (10.12+) or Linux, you can download binary releases here.

Homebrew

brew tap mingeme/tap
brew install fuckmit

Install from crates.io

cargo install fuckmit

Install from Source

If you have the Rust toolchain installed (including cargo), you can install directly from the GitHub repository:

cargo install --locked --git https://github.com/mingeme/fuckmit

Or clone and build manually:

# Clone repository
git clone https://github.com/mingeme/fuckmit.git
cd fuckmit

# Build project
cargo build --release

# Install binary
cargo install --path .

Supported AI Providers

Provider	Status	Supported Models
OpenAI	✅	GPT-3.5, GPT-4, etc
Azure OpenAI	✅	GPT-3.5, GPT-4, etc
DeepSeek	✅	DeepSeek Chat
Qwen	✅	Qwen Turbo, etc

Environment Configuration

Environment Variables

OpenAI

export OPENAI_API_KEY="your-openai-api-key"
export OPENAI_MODEL="gpt-4"  # Optional, defaults to gpt-3.5-turbo
export OPENAI_BASE_URL="https://api.openai.com/v1"  # Optional

Azure OpenAI

export AZURE_OPENAI_API_KEY="your-azure-api-key"
export AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com"
export AZURE_OPENAI_DEPLOYMENT="your-deployment-name"
export AZURE_OPENAI_API_VERSION="2024-02-15-preview"  # Optional

DeepSeek

export DEEPSEEK_API_KEY="your-deepseek-api-key"
export DEEPSEEK_MODEL="deepseek-chat"  # Optional
export DEEPSEEK_BASE_URL="https://api.deepseek.com/v1"  # Optional

Qwen

export QWEN_API_KEY="your-qwen-api-key"
export QWEN_MODEL="qwen-turbo"  # Optional
export QWEN_BASE_URL="https://dashscope.aliyuncs.com/compatible-mode/v1"  # Optional

Global Settings

export LLM_MODEL="deepseek/deepseek-chat"  # Required
export LLM_TIMEOUT_SECONDS="30"  # Optional, timeout setting
export LLM_MAX_RETRIES="3"  # Optional, retry count

Commit Message Rules

export FUCKMIT_RULES="Use English commit messages"  # Optional, custom rules for commit message generation

Note: The --rules CLI flag takes precedence over the FUCKMIT_RULES environment variable.

Interactive Mode

export FUCKMIT_INTERACTIVE="true"  # Optional, enable interactive mode by default
export FUCKMIT_CANDIDATES="3"  # Optional, number of candidates to generate (1-5, default: 3)

Usage

Basic Usage

# Generate and commit Git commit message
fuckmit

# Only display generated commit message, don't actually commit
fuckmit --dry-run

# Use provider/model format
fuckmit --model openai/gpt-4

# Add custom rules via CLI flag
fuckmit --rules "Use English commit messages"

# Or set rules via environment variable (applies to all invocations)
export FUCKMIT_RULES="Use English commit messages"
fuckmit

# Add change context
fuckmit --context "Fixed user login bug"

# Use rules and context together
fuckmit --rules "Use concise descriptions" --context "Refactored database connection logic"

# Custom AI parameters
fuckmit --max-tokens 1000 --temperature 0.5

Interactive Mode

Interactive mode allows you to select from multiple AI-generated commit message candidates and edit them before committing.

# Enable interactive mode with default 3 candidates
fuckmit --interactive

# Generate 5 candidates to choose from
fuckmit --interactive --candidates 5

# Use environment variable to enable by default
export FUCKMIT_INTERACTIVE=true
fuckmit

Interactive Mode Features:

• Generate multiple commit message candidates in parallel
• Select from a numbered list of options
• Edit selected message in your default editor ($EDITOR or $VISUAL)
• Quick edit mode for type/scope/description
• Automatic deduplication of identical messages
• Graceful fallback to non-interactive mode in non-TTY environments

Command Line Arguments

• -d, --dry-run: Only display generated commit message, don't execute commit
• -i, --interactive: Enable interactive mode to select from multiple candidates
• --candidates <NUM>: Number of candidates to generate in interactive mode (1-5, default: 3)
• -m, --model <MODEL>: Specify AI model or use "provider/model" format
• -r, --rules <RULES>: Custom commit message generation rules
• -c, --context <CONTEXT>: Provide additional context for changes
• --max-tokens <NUM>: Maximum tokens for generated message (default: 8192)
• --temperature <NUM>: AI generation temperature parameter, range 0.0-2.0 (default: 0.7)

License

This project is open source under the MIT License - see the LICENSE file for details.