GitMind: AI Commit Assistant for VS Code

Professional AI-powered commit message generation for Visual Studio Code. Leverage 13 different AI providers including OpenAI GPT-4o, Claude-3-5-sonnet, Gemini-2.5-flash, DeepSeek-reasoner, Grok-3, Perplexity-sonar, and 50+ models to create consistent, conventional commit messages that improve code history quality and team collaboration.

Installation

Install from the VS Code Marketplace or via Quick Open (Ctrl+P):

ext install ShahabBahreiniJangjoo.ai-commit-assistant

Core Features

Multi-Provider AI Support

Access 13 different AI providers with unified configuration and intelligent fallback handling. From zero-setup GitHub Copilot to privacy-focused local Ollama deployments supporting GPT-4o, Claude-opus-4, Gemini-2.5-pro, DeepSeek-chat, Grok-3, Perplexity-sonar, Mistral-large, and 50+ additional models.

Advanced Git Integration

Smart diff analysis with automatic staging detection, binary file handling, and comprehensive repository state management. Handles complex scenarios including merge conflicts and mixed changes.

Conventional Commits Standard

Intelligent Diff Processing

Enhanced change detection for staged and unstaged files with context-aware analysis and user guidance for complex scenarios.

Dynamic Model Selection

Real-time model browsing for compatible providers with search and filtering capabilities. 2M+ token context windows supported for large repositories with models like Gemini-2.5-pro and Claude-3-5-sonnet.

Smart Prompt Management

Save Last Custom Prompt feature allows you to automatically save and reuse custom prompts across commit generations. Saved prompts appear as defaults in future sessions with clipboard copy options for easy editing.

Professional Workflow Integration

Native VS Code SCM panel integration with loading indicators, comprehensive status feedback, and standardized prompt engineering across all providers.

AI Provider Ecosystem

Model Comparison & Selection Guide

Provider	Featured Models	Context	Free Tier	Setup	Strengths	Limitations
GitHub Copilot	gpt-4o, claude-3.5-sonnet, o3	128k	No	5sec	Zero config, VS Code native	Requires subscription
Google Gemini	2.5-flash, 2.5-pro, 2.0-flash	2M	15 RPM	2min	Massive context, thinking model	Rate limited (free)
Grok (X.ai)	grok-3, grok-3-fast, grok-3-mini	128k	Limited	2min	Real-time data access, fast	X Premium required
DeepSeek	reasoner, chat	128k	50 RPM	2min	Advanced reasoning, cost-effective	Newer provider
Perplexity	sonar-pro, sonar-reasoning, sonar	127k	Limited	2min	Real-time web search, reasoning	Paid tier recommended
Mistral AI	large-latest, medium, small	128k	1 RPM	2min	EU-compliant, multilingual	Low free tier limits
Ollama	deepseek-r1, llama3.3, phi4, qwen3	128k	Unlimited	5min	Complete privacy, no API costs	Hardware dependent
OpenAI	gpt-4o, gpt-4.1, o3, o4-mini	128k	No	2min	Industry standard, multimodal	Paid only, higher cost
Anthropic	claude-opus-4, sonnet-4, haiku	200k	No	2min	Superior reasoning, long context	Paid only
Together AI	Llama-3.3-70B, Mixtral-8x7B	128k	60 RPM	2min	Optimized inference, generous free	Variable model quality
Hugging Face	Mistral-7B, Zephyr-7B, OpenHermes	32k	Varies	2min	Open source, customizable	Inconsistent performance
Cohere	command-r, command-a-03-2025	128k	20 RPM	2min	RAG-optimized, retrieval focus	Limited model variety
OpenRouter	Multiple providers & models	Varies	Limited	2min	Access to premium models	Complex pricing

Quick Selection Guide

For Immediate Use: GitHub Copilot with gpt-4o or claude-3.5-sonnet (existing subscription) or Google Gemini 2.5-flash (best free tier)

For Privacy: Ollama with local phi4, llama3.3:70b, or codellama deployment

For Performance: OpenAI GPT-4.1, Anthropic Claude-opus-4, or DeepSeek-reasoner

For Large Projects: Google Gemini-2.5-pro (2M context) or Anthropic Claude-sonnet-4 (200k context)

API Provider Setup Links

Get API Keys & Documentation

Provider	API Signup	Documentation	Free Tier Details
GitHub Copilot	VS Code Copilot	Copilot Docs	Requires subscription
Google Gemini	AI Studio	Gemini API Docs	15 requests/minute
Grok (X.ai)	X.ai Console	Grok API Docs	Limited free tier
DeepSeek	DeepSeek Platform	DeepSeek API Docs	50 requests/minute
Perplexity	Perplexity Settings	Perplexity API Docs	Limited free tier
Mistral AI	Mistral Console	Mistral API Docs	1 request/minute
OpenAI	OpenAI Platform	OpenAI API Docs	Paid only
Anthropic	Anthropic Console	Claude API Docs	Paid only
Together AI	Together Platform	Together API Docs	60 requests/minute
Hugging Face	HF Token	HF Inference API	Varies by model
Cohere	Cohere Dashboard	Cohere API Docs	20 requests/minute
OpenRouter	OpenRouter Keys	OpenRouter Docs	Limited free tier
Ollama	Ollama Download	Ollama Docs	Unlimited (local)

Quick API Setup Guide

Choose your provider from the table above
Sign up using the provided API signup link
Generate API key in your provider dashboard
Configure GitMind with your API key
Select optimal model for your use case

For detailed setup instructions for each provider, see our Configuration Guide.

Supported Models by Provider

Complete Model Directory (50+ Models Supported)

GitHub Copilot Models

Google Gemini Models

Grok (X.ai) Models

OpenAI Models

Anthropic Claude Models

DeepSeek Models

deepseek-chat | deepseek-reasoner

Perplexity Models

Mistral AI Models

Cohere Models

Together AI Models

meta-llama/Llama-3.3-70B-Instruct-Turbo | meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo | meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo | meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo | microsoft/WizardLM-2-8x22B | NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO | deepseek-ai/deepseek-llm-67b-chat | deepseek-ai/DeepSeek-R1-Distill-Llama-70B-free

Hugging Face Models

Ollama Local Models

OpenRouter Models

Access to multiple providers including google/gemma-3-27b-it:free and premium models from various providers

Technical Architecture

Advanced Diff Analysis Engine

Smart Staging Detection: Automatic detection of staged vs unstaged changes with user guidance
Binary File Handling: Proper detection and exclusion of binary files from analysis
Repository State Management: Comprehensive handling of complex git repository states
Edge Case Processing: Robust handling of empty diffs, merge conflicts, and repository anomalies

Commit Message Intelligence

Context Analysis: Deep analysis of file diffs, change patterns, and repository history powered by advanced models like GPT-4o, Claude-opus-4, and Gemini-2.5-pro
Scope Detection: Automatic identification of affected modules and components
Breaking Change Recognition: Intelligent detection of API changes and breaking modifications
Verbosity Control: Toggle between detailed descriptions and concise summaries

Enterprise-Ready Configuration

Standardized Prompts: Consistent prompt templates across all 12 AI providers and 50+ models
Custom Context Enhancement: Optional domain-specific prompt customization with automatic saving and reuse
Smart Prompt Management: Save Last Custom Prompt feature for workflow efficiency and consistency
Token Optimization: Smart content truncation and cost management with pre-generation estimation
Rate Limit Management: Advanced monitoring with minute-based tracking and anomaly detection
Debug Mode: Comprehensive API interaction logging and response analysis

Quick Start Guide

1. Choose Your Provider Strategy

For Immediate Use: GitHub Copilot with gpt-4o or claude-3.5-sonnet (if you have VS Code Copilot subscription)
For Free Usage: Google Gemini-2.5-flash or DeepSeek-reasoner for best free tier experience
For Privacy: Ollama with local deepseek-r1, llama3.3, phi4, qwen3, gemma3, or codellama deployment
For Performance: OpenAI gpt-4.1, Anthropic claude-opus-4, or Google gemini-2.5-pro

2. Configuration

Open VS Code Source Control panel
Click the settings icon in GitMind section
Select your preferred AI provider
Add API key (skip for GitHub Copilot and Ollama)
Choose optimal model for your use case
Optional: Enable "Prompt Customization" and "Save Last Custom Prompt" for enhanced workflow

3. Generate Commits

Stage your changes in Git
Click the "AI" button in Source Control panel
Optional: Add custom context (saved automatically if enabled)
Review and edit the generated commit message
Commit your changes

Model Selection Guidelines

Context Window Considerations

Large repositories/refactoring: 128k+ tokens (Gemini-2.5-pro, Claude-opus-4, GPT-4o)
Standard commits: 32k tokens sufficient (Mistral-medium, most models)
Complex multi-file changes: 2M tokens recommended (Gemini-2.5-pro models)

Performance Optimization

Speed priority: Gemini-2.0-flash, Mistral-small, Command-light, GPT-4o-mini
Quality priority: GPT-4.1, Claude-opus-4, Llama-3.3-70B, DeepSeek-reasoner
Cost optimization: Hugging Face open models, Ollama local deployment

Specialized Use Cases

Code-heavy changes: deepseek-r1, codellama:34b (Ollama), DeepSeek models, Qwen2.5-coder, starcoder2
Multilingual projects: Mistral models, Gemini series, aya-expanse models
Documentation changes: Command-light (Cohere), smaller efficient models

Advanced Features

Enhanced Repository Analysis

Intelligent change categorization with automatic detection of feature additions, bug fixes, documentation updates, and refactoring patterns using advanced models like Claude-opus-4 and GPT-4.1.

Smart Prompt Management System

Comprehensive prompt lifecycle management with automatic saving, intelligent reuse, and seamless workflow integration:

Automatic Saving: Custom prompts are automatically saved when "Save Last Custom Prompt" is enabled
Smart Defaults: Saved prompts appear as default values in future commit generations
Clipboard Integration: Built-in copy-to-clipboard functionality for external editing
Management Commands: Dedicated VS Code commands for viewing and clearing saved prompts
Persistent Storage: Prompts persist across VS Code sessions and workspaces using global configuration

Unified Prompt Architecture

Standardized prompt engineering across all 12 providers and 50+ models ensures consistent, high-quality commit messages regardless of chosen AI model.

Advanced Prompt Management

Save Last Custom Prompt: Automatically save and reuse custom prompts across sessions
Smart Defaults: Saved prompts appear as default values with clipboard copy options
Prompt Commands: Dedicated commands for viewing (View Last Custom Prompt) and clearing (Clear Last Custom Prompt) saved prompts
Persistent Storage: Prompts persist across VS Code sessions and workspaces

Professional Integration

Native VS Code Source Control panel integration
Batch processing support for multiple file changes
Manual override with edit preservation
Comprehensive error handling with actionable guidance

Diagnostic & Monitoring Tools

Real-time token usage estimation
API response analysis and debugging
Rate limit monitoring and optimization
Model performance analytics

Configuration Options

Access settings via Command Palette: AI Commit Assistant: Open Settings or the settings icon in the Source Control panel.

Provider Management

AI provider selection with real-time validation
Secure API key configuration
Model selection with performance metrics

Prompt Customization

Enable custom context prompts for commit generation
Save Last Custom Prompt toggle for automatic prompt reuse
Persistent prompt storage across sessions

Message Formatting

Conventional commit standard compliance
Verbosity level control
Custom scope and type configuration

Advanced Settings

Debug mode for development
Custom prompt templates
Token usage optimization
Rate limit configuration

Requirements

Visual Studio Code ^1.100.0
Git repository (initialized)
API key from chosen provider OR Ollama installation for local deployment

Professional Use Cases

Individual Developers

Maintain consistent commit history with intelligent message generation that adapts to your coding patterns and project context using models like GPT-4o, Claude-3.5-sonnet, and Gemini-2.5-flash.

Team Collaboration

Standardize commit message formats across team members with configurable conventional commit standards and custom prompt templates powered by enterprise-grade models.

Enterprise Deployment

Scale across organizations with support for multiple AI providers, usage analytics, and centralized configuration management using premium models like Claude-opus-4 and GPT-4.1.

Open Source Projects

Leverage free tier providers (Gemini-2.5-flash, DeepSeek-reasoner) or local Ollama deployment (deepseek-r1, phi4, llama3.3, qwen3, gemma3) to maintain professional commit standards without API costs.

GitMind transforms your development workflow with intelligent, context-aware commit message generation. Supporting 11 AI providers with 50+ models including GPT-4o, Claude-opus-4, Gemini-2.5-pro, DeepSeek-R1, Llama-3.3, Phi-4 and advanced diff analysis, it delivers professional-grade commit messages that improve code history quality and team collaboration efficiency.

Development and Testing

GitMind includes a comprehensive automated test suite to ensure reliability and quality before publication.

Test Coverage

The extension features 7 comprehensive test suites covering all main functionality:

Settings UI Testing: UI persistence, validation, and provider switching
AI Providers Testing: All 13 providers with API validation and model selection
Extension Commands Testing: All commands, error handling, and status management
Git Integration Testing: Diff parsing, commit messages, and repository validation
Webview Components Testing: Settings panel, onboarding, and messaging
Error Handling Testing: API errors, network issues, rate limits, and recovery
Configuration Management Testing: Settings persistence, schema validation, and migration

Quality Assurance

✅ 100% TypeScript compilation with no errors or warnings
✅ Complete VS Code API mocking for isolated testing
✅ Comprehensive error simulation for robust error handling
✅ End-to-end integration testing for all workflows
✅ Modular test architecture for maintainability

Running Tests

# Compile and run all tests
npm test

# Check test status and coverage
npm run test:status

# Validate test compilation
npm run test:validate

For detailed test documentation, see TEST_DOCUMENTATION.md.

Privacy and Telemetry

GitMind collects anonymous usage data to help improve the extension. This helps us understand how the extension is used and identify areas for improvement.

What data is collected:

Usage Analytics: Command usage frequency, success/failure rates, and performance metrics
Provider Statistics: Which AI providers are used (but not API keys or responses)
Technical Information: VS Code version, OS platform, extension version
Error Reports: Anonymous error logs and exception details
User Flow: Navigation patterns within the extension

What is NOT collected:

No code content: Your actual code, commit messages, or diff content
No personal information: Names, emails, or other personal identifiers
No API keys: Your API credentials are never transmitted
No repository information: Project names, file paths, or repository details

How to disable telemetry:

You can disable telemetry at any time by:

Opening VS Code Settings (Ctrl/Cmd + ,)
Searching for "telemetry"
Setting "Telemetry: Telemetry Level" to "off"

GitMind respects your privacy settings and will not collect any data if telemetry is disabled.

Data usage:

The collected data helps us:

Improve extension reliability and performance
Understand which features are most valuable
Prioritize development efforts
Fix bugs and compatibility issues

All data is processed in accordance with Microsoft's privacy policies and is used solely for improving the GitMind extension.

Name		Name	Last commit message	Last commit date
Latest commit History 337 Commits
.github/ISSUE_TEMPLATE		.github/ISSUE_TEMPLATE
azure-deployment		azure-deployment
images		images
instructions		instructions
resources		resources
scripts		scripts
src		src
.env.azure		.env.azure
.env.example		.env.example
.gitignore		.gitignore
.npmrc		.npmrc
.vscode-test.mjs		.vscode-test.mjs
.vscodeignore		.vscodeignore
CHANGELOG.md		CHANGELOG.md
LICENSE.md		LICENSE.md
README.md		README.md
esbuild.js		esbuild.js
eslint.config.mjs		eslint.config.mjs
package.json		package.json
tsconfig.json		tsconfig.json

License

shahabahreini/AI-Commit-Assistant

Folders and files

Latest commit

History

Repository files navigation