OmniDev AI Agent

🚀 OmniDev

Your Multi-Model AI Development Assistant - Free, Intelligent, Autonomous

OmniDev is an intelligent CLI-based AI coding assistant that brings the power of multiple AI models to your terminal - with a free tier. It combines autonomous operation, strategic planning, and smart model selection to help you code faster and better.

---

✨ What OmniDev Does

OmniDev is your AI pair programmer that:

• 🤖 Creates, edits, and manages files automatically based on natural language instructions
• 🧠 Plans complex changes before executing them, showing you the full impact
• 🎯 Intelligently selects the best AI model for each task (coding, debugging, refactoring, testing)
• 🔄 Maintains dynamic project context - automatically includes relevant files without manual selection
• 💰 Free tier available using Groq (30 requests/minute), with optional premium API upgrades
• 🛡️ Keeps your code safe with automatic backups and Git integration
• 🎨 Modern Rich CLI with colorful UI, loading spinners, slash commands, and interactive REPL mode
• 🤝 CrewAI Agent System for intelligent orchestration of all internal operations
• ⚙️ Project-specific configuration with .env file support for API keys

---

🎮 Four Operational Modes

1. Agent Mode 🤖

Full autonomy - tell it what you want, and it handles everything automatically.

$ omnidev "Build a REST API for user authentication with JWT"

→ Creating project structure...
✓ Created 5 files
✓ Implemented authentication logic
✓ Generated tests
✓ All tests passing

Done! Your API is ready.

2. Planning Mode 📋

Strategic approach - shows you the plan first, then executes with your approval.

$ omnidev --mode planning "Refactor authentication to use OAuth2"

→ Creating refactoring plan...

PLAN: OAuth2 Migration
├─ Phase 1: Setup (5 files, ~15 min)
├─ Phase 2: Core migration (4 files, ~20 min)  
├─ Phase 3: Integration (3 files, ~10 min)
└─ Phase 4: Testing (6 files, ~10 min)

Proceed? (yes/no/modify):

3. Auto-Select Mode 🎯

Smart model routing - automatically picks the best AI model for each task.

$ omnidev "Fix this performance issue"

🎯 Selected: Claude Sonnet 4 (best for optimization)
→ Analyzing code...
→ Found bottleneck in database query
✓ Applied optimization (35x faster)

4. Manual Mode 🎮

Full control - you approve every step and choose which AI model to use.

$ omnidev --mode manual "Create a new component"

Which model? (gpt-4o/claude/deepseek): gpt-4o
Should I create new file? (yes/no): yes
Filename: components/UserProfile.tsx
✓ Created

---

🚀 Quick Start

Installation

For Users:

# Install OmniDev
pip install git+https://github.com/codewithdark-git/OmniDev.git

# Run interactive setup wizard
omnidev -i

For Developers: See README_SETUP.md for detailed setup instructions using Miniconda and UV package manager.

Interactive Mode (Recommended)

Start OmniDev in interactive REPL mode for a chat-like experience:

omnidev -i
# or
omnidev --interactive

This launches an interactive shell with:

• 💬 Chat-style interface
• 🔄 Loading spinners while waiting for responses
• ⚙️ In-REPL setup wizard (/setup)
• 🎨 Beautiful terminal UI with panels and colors
• 📜 Command history and auto-suggestions
• 🔧 Slash commands for quick actions

Available Slash Commands:

Command	Description
/help	Show all available commands
/setup	Run full setup wizard
/provider	Change AI provider
/model	Change AI model
/mode	Change operational mode
/status	Show current configuration
/clear	Clear the screen
/exit	Exit the REPL

Provider Setup

OmniDev supports multiple AI providers. On first run, you'll be guided through setup:

Supported Providers:

Provider	Free Tier	API Key Required
Groq	✅ Yes (30 req/min)	✅ Yes (free at console.groq.com)
OpenAI	❌ No	✅ Yes
Anthropic	❌ No	✅ Yes
Google	❌ No	✅ Yes
OpenRouter	✅ Yes	✅ Yes

Quick Setup:

# Interactive setup (recommended)
omnidev -i
# Then type: /setup

# Or run setup directly
omnidev --setup

Manual Setup: Create a .env file in your project root:

# .env
GROQ_API_KEY=your-groq-api-key-here
# or
OPENAI_API_KEY=your-openai-api-key-here

Important: Add .env to your .gitignore to keep your API keys secure.

Basic Usage

# Start interactive mode (recommended)
omnidev -i

# Run a single query
omnidev "create a Python FastAPI server with authentication"

# Agent mode (full autonomy)
omnidev --mode agent "add user registration endpoint"

# Planning mode (review before execution)
omnidev --mode planning "refactor the database layer"

# Use specific model
omnidev --model gpt-4o "explain how this algorithm works"

---

🎯 Core Features

Intelligent File Operations

• Create new files with proper structure and conventions
• Edit existing files with surgical precision
• Delete files safely with confirmation
• Multi-file coordination for complex refactoring

Dynamic Context Management

• Automatically includes relevant files based on your task
• Learns which files you work on together
• Optimizes token usage to fit more context
• Updates context in real-time as you work

Smart Model Routing

• Analyzes each task and selects the optimal AI model
• Balances quality, speed, and cost automatically
• No automatic fallback - shows clear errors when provider fails
• Respects your provider/model selection

Safety & Reliability

• Automatic backups before any destructive operation
• Git integration with smart commit messages
• Easy rollback with one command
• Validates code before applying changes

---

💡 Example Workflows

Create a New Feature

$ omnidev "add password reset functionality to the auth system"

→ Planning implementation...
→ Creating email templates...
→ Adding reset token logic...
→ Updating API endpoints...
→ Generating tests...
✓ Feature complete! 8 files modified, 247 lines added

Debug an Issue

$ omnidev "why is the /users endpoint returning 500 errors?"

🎯 Selected: GPT-4 Turbo (best for debugging)
→ Analyzing error logs...
→ Found: Missing database migration
→ Solution: Run migration 'add_user_email_index'

Apply fix? (yes/no): yes
✓ Migration applied
✓ Tests passing

Interactive Session

$ omnidev -i

╭────────────────────────────────────────────╮
│  Welcome to OmniDev Interactive Mode       │
│                                            │
│    ⚡ Provider: groq                       │
│    🤖 Model:    llama-3.3-70b-versatile    │
│    📋 Mode:     agent                      │
│                                            │
│    Type /help for commands!                │
╰────────────────────────────────────────────╯

❯ tell me about the current project structure

💭 Thinking with llama-3.3-70b-versatile via Groq (Free)...

╭────────────────────────────────────────────╮
│  The project follows a modular structure   │
│  with clear separation of concerns...      │
╰────────────────────────────────────────────╯

---

🔧 Configuration

OmniDev works out of the box, but you can customize it:

Interactive Setup (Recommended)

# Start interactive mode and run setup
omnidev -i
❯ /setup

Global Configuration

# Set your preferred default model
omnidev config set default-model llama-3.3-70b-versatile

# Add your API keys
omnidev config add-key groq YOUR_API_KEY
omnidev config add-key openai YOUR_API_KEY
omnidev config add-key anthropic YOUR_API_KEY

Project Configuration

Create .omnidev.yaml in your project root:

project_name: "My API Project"

mode:
  default_mode: agent

models:
  preferred: llama-3.3-70b-versatile
  fallback: groq

context:
  always_include:
    - "config/*.py"
    - "models/*.py"
  exclude:
    - "*.log"
    - "node_modules/*"

---

🌟 Why OmniDev?

vs. Claude Code

• ✅ Free tier available (Groq offers 30 req/min free)
• ✅ Multiple AI models (not locked to one provider)
• ✅ Smart model selection (uses best model for each task)
• ✅ Planning mode (see changes before they happen)

vs. Cursor

• ✅ CLI-first design (works with any editor)
• ✅ Autonomous modes (less manual intervention)
• ✅ Free to start (no subscription required)
• ✅ Open source (customize and extend)

vs. GitHub Copilot

• ✅ Full file operations (not just autocomplete)
• ✅ Project-wide context (understands your entire codebase)
• ✅ Multi-model support (not limited to OpenAI)
• ✅ Intelligent planning (thinks before acting)

---

🛠️ Technology

Built with:

• Python 3.10+ for core logic
• Groq for fast, free AI inference
• Official APIs: OpenAI, Anthropic, Google, OpenRouter (optional)
• Rich for beautiful terminal interface
• prompt_toolkit for interactive REPL
• GitPython for version control integration
• CrewAI for agent orchestration

Supported AI Models:

• LLaMA 3.3 70B, LLaMA 3.1 8B, Mixtral (Groq - Free)
• GPT-4o, GPT-4 Turbo, GPT-4o-mini (OpenAI)
• Claude Sonnet 4, Claude Opus 4 (Anthropic)
• Gemini 2.0 Flash, Gemini 2.5 Pro (Google)
• Many more through OpenRouter

Supported Languages & Frameworks:

• Python (Django, FastAPI, Flask)
• JavaScript/TypeScript (React, Node.js, Express, Next.js)
• Go, Rust, Java, C#, PHP, Ruby
• And more...

---

📋 Requirements

• Python 3.10 or higher
• Git (for version control features)
• Internet connection (for AI models)

Required:

• API key from at least one provider (Groq is free!)

Optional (for premium models):

• OpenAI API key (for GPT-4, GPT-4o)
• Anthropic API key (for Claude models)
• Google API key (for Gemini models)
• OpenRouter API key (for access to many models)

---

🗺️ Roadmap

✅ Current (v0.1 - MVP)

• Agent, Planning, Auto-Select, and Manual modes
• File create, edit, delete operations
• Groq integration (free tier)
• Interactive REPL mode with loading spinners
• Basic context management
• Git integration
• In-REPL setup wizard

🚧 In Progress (v0.2)

• Web search integration
• Documentation fetching
• Enhanced context scoring
• Streaming responses

📅 Coming Soon (v0.3+)

• MCP server support
• IDE integrations (VS Code, JetBrains)
• Advanced testing capabilities
• Performance profiling
• Team collaboration features

---

🤝 Contributing

We welcome contributions! Whether it's:

• 🐛 Bug reports
• 💡 Feature suggestions
• 📝 Documentation improvements
• 🔧 Code contributions

Please see CONTRIBUTING.md for guidelines.

---

📄 License

MIT License - see LICENSE for details.

---

🙏 Acknowledgments

• Groq for providing fast, free AI inference
• Rich for the beautiful terminal UI
• CrewAI for agent orchestration
• The open-source community for inspiration and support

---

💬 Community & Support

• GitHub Issues: Report bugs or request features
• Discussions: Ask questions and share ideas
• Discord: Join our community (coming soon)
• Twitter: @omnidev_ai (coming soon)

---

⚡ Quick Examples

# Start interactive mode (recommended)
omnidev -i

# Create a complete web application
omnidev "build a todo app with React frontend and FastAPI backend"

# Debug production issues
omnidev "analyze why the server is slow and fix it"

# Refactor legacy code
omnidev --mode planning "modernize this codebase to use async/await"

# Learn from your code
omnidev "explain how the authentication flow works"

# Generate documentation
omnidev "create API documentation for all endpoints"

# Add new features
omnidev "add user profile editing with avatar upload"

---

<div align="center">

Made with ❤️ for developers who want AI assistance without the premium price tag

⭐ Star us on GitHub • 📖 Read the Docs • 🐦 Follow Updates

</div>