GitMind: AI Commit Assistant for VS Code

Generate meaningful and consistent git commit messages using AI. This extension leverages multiple AI providers, focusing on those offering free tier API tokens, to help you create conventional commit messages that are clear, consistent, and informative.

GitMind Demo

Installation

Install directly from the VS Code Marketplace

Or launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter: ext install ShahabBahreiniJangjoo.ai-commit-assistant

Features

Multiple AI Provider Support

Google's Gemini AI (Recommended): Advanced language model for precise commit messages
Mistral AI: High-performance language model with multiple tiers, including free options
Hugging Face: Access to various open-source language models with free tier options
Ollama: A completely free, locally installed engine for privacy-focused development
Cohere: High-quality language models with a focus on command and retrieval tasks, offering various models and a free tier
OpenAI: High-quality GPT models for professional results
Together AI: Platform for accessing high-quality open-source models with free tier options
OpenRouter: Gateway to multiple AI models from different providers with a unified API

Provider	Free Plan	Paid Plan
Google's Gemini AI	Yes	Yes
Mistral AI	Yes	Yes
Hugging Face	Yes	Yes
Ollama	Yes	No
Cohere	Yes	Yes
OpenAI	No	Yes
Together AI	Yes	Yes

Smart Git Integration

Automatic Change Detection:
- Analyzes staged changes automatically
- Handles unstaged changes with user confirmation
- Supports both single and multiple file changes
Custom Context Support:
- Optional prompt customization for better context
- User-provided additional information for commit messages
- Enhanced commit message accuracy with custom context
VS Code SCM Integration:
- AI icon in Source Control panel for quick access
- ⚙️ Settings icon for quick configuration
- 🔄 Loading indicator during message generation
Flexible Workflow Support:
- Works with staged and unstaged changes
- Compatible with all Git operations
- Preserves manual edits
Verbosity Control:
- Toggle between detailed and concise commit messages
- Detailed mode includes bullet points and context
- Concise mode shows only the summary line
Diagnostic Information:
- Optional token count estimation
- Model information display
- Generation status in status bar

Commit Message Generation

Conventional Commits Format:
- Type categorization (feat|fix|docs|style|refactor|test|chore)
- Scope support (optional)
- Breaking change detection
Smart Formatting:
- Subject line limited to 72 characters
- Detailed bullet-point descriptions
- Proper imperative mood
- Technical accuracy

Token Management

Token Estimation:
- Pre-generation token count estimates
- Smart content optimization
- Rate limit monitoring (Mistral)
Usage Tracking:
- Real-time token consumption display
- Rate limit status indicators
- Quota management helpers

Prompt Customization

Enable prompt customization to provide additional context when generating commit messages:

Enable via Settings:
- Open VS Code settings
- Search for "AI Commit Assistant"
- Enable "Prompt Customization"
Usage:
- When generating a commit message, you'll be prompted for additional context
- Enter relevant information about the changes
- The AI will incorporate your context into the commit message

Advanced Features

Debug Mode:
- Detailed API interaction logs
- Response tracking
- Error diagnostics
Rate Limit Handling (only Mistral):
- Smart API quota management
- Clear feedback on limits
- Automatic retries (where applicable)

Requirements

Visual Studio Code ^1.75.0
Git installed and initialized
One of the following configurations:
- Gemini API key (free tier available)
- Hugging Face API key (free tier available)
- Ollama installation (completely free)
- Mistral API key (free tier available)
- Cohere API key (free tier available)

Configuration

Via Settings UI

Access the settings UI through:

Command Palette: AI Commit Assistant: Open Settings
⚙️ icon in Source Control panel
Verbose commit message toggle
Diagnostic information toggle
Provider-specific model selection
Real-time validation
Direct links to API documentation

AI Provider Setup

Google's Gemini AI

Sign up for a Google Cloud account at Google AI Studio
Create an API key in your Google AI Studio dashboard
Configure the extension:
- Set "Gemini" as your provider
- Enter your API key
- Select your preferred model (gemini-pro)

Mistral AI

Sign up for an account at Mistral AI
Create an API key in your Mistral AI dashboard
Configure the extension:
- Set "Mistral" as your provider
- Enter your API key
- Select your preferred model (mistral-large-latest, mistral-medium, etc.)

Hugging Face

Sign up for an account at Hugging Face
Create an API key in your Hugging Face dashboard
Configure the extension:
- Set "Hugging Face" as your provider
- Enter your API key
- Select or enter your preferred model identifier

Ollama

Install Ollama on your local machine
Pull your preferred model (e.g., ollama pull phi4)
Configure the extension:
- Set "Ollama" as your provider
- Configure the Ollama URL (default is http://localhost:11434)
- Select your preferred model from those you've pulled

Cohere

Sign up for an account at Cohere
Create an API key in your Cohere dashboard
Configure the extension:
- Set "Cohere" as your provider
- Enter your API key
- Select your preferred model (command, command-light, command-nightly, or command-r)

OpenAI

Sign up for an account at OpenAI
Create an API key in your OpenAI dashboard
Configure the extension:
- Set "OpenAI" as your provider
- Enter your API key
- Select your preferred model (GPT-4o, GPT-4-turbo, or GPT-3.5-turbo)

Together AI

Sign up for an account at Together AI
Create an API key in your Together AI dashboard
Configure the extension:
- Set "Together" as your provider
- Enter your API key
- Select your preferred model (e.g., meta-llama/Llama-3.3-70B-Instruct-Turbo)

OpenRouter

Sign up for an account at OpenRouter
Create an API key in your OpenRouter dashboard
Configure the extension:
- Set "OpenRouter" as your provider
- Enter your API key
- Select your preferred model (e.g., google/gemma-3-27b-it:free, openai/gpt-4-turbo)

Available Models

Gemini Models

gemini-pro: Advanced language model with strong reasoning capabilities and code understanding

Mistral Models

mistral-large-latest: Most powerful Mistral model with excellent reasoning capabilities
mistral-medium: Balanced model offering good performance at lower cost
mistral-small: Lightweight model for basic tasks with minimal token usage

Hugging Face Models

mistralai/Mistral-7B-Instruct-v0.3: State-of-the-art open-source 7B parameter model
meta-llama/Llama-2-7b-chat-hf: Meta's Llama 2 conversational model
google/flan-t5-xxl: Google's instruction-tuned T5 model
And many other open-source models available on Hugging Face

Ollama Models

phi4: Microsoft's advanced model with excellent reasoning capabilities
llama3: Meta's latest Llama model
mistral: Mistral AI's 7B model running locally
codellama: Specialized for code understanding and generation

Cohere Models

command: Optimized for instructions with excellent reasoning capabilities
command-light: Lighter, faster version of command with lower cost
command-nightly: Latest build with newest improvements (may be less stable)
command-r: Optimized for retrieval and RAG applications

OpenAI Models

gpt-4o: Latest GPT-4o model with enhanced capabilities
gpt-4-turbo: High-performance GPT-4 model with fast inference
gpt-3.5-turbo: Efficient and cost-effective GPT-3.5 model

Together AI Models

meta-llama/Llama-3.3-70B-Instruct-Turbo: Meta's powerful Llama 3.3 model optimized for instruction following
mistralai/Mixtral-8x7B-Instruct-v0.1: Mistral's high-performing mixture-of-experts model
meta-llama/Llama-2-70b-chat-hf: Meta's large Llama 2 model for chat applications
And many other open-source models available on the Together AI platform

GitMind: AI Commit Assistant

Shahab Bahreini Jangjoo