██████╗ ██╗ ██████╗ ██╗ ██╗████████╗███╗ ██╗ ██████╗ ██╗ ██╗
██╔══██╗██║██╔════╝ ██║ ██║╚══██╔══╝████╗ ██║██╔═══██╗██║ ██║
██████╔╝██║██║ ███╗███████║ ██║ ██╔██╗ ██║██║ ██║██║ █╗ ██║
██╔══██╗██║██║ ██║██╔══██║ ██║ ██║╚██╗██║██║ ██║██║███╗██║
██║ ██║██║╚██████╔╝██║ ██║ ██║ ██║ ╚████║╚██████╔╝╚███╔███╔╝
╚═╝ ╚═╝╚═╝ ╚═════╝ ╚═╝ ╚═╝ ╚═╝ ╚═╝ ╚═══╝ ╚═════╝ ╚══╝╚══╝
CUDA AI Assistant • Open Source
rightnowai.co • @rightnowai_co • Discord
Open Source CLI Tool - Want a full-featured CUDA development environment with code completion, integrated debugging, and advanced AI features? Try RightNow Code Editor - Our complete IDE for GPU development.
RightNow CLI is an AI-powered CUDA development assistant that helps you write, optimize, and debug GPU code. Start FREE with no credit card required!
pip install rightnow-cli
Linux/macOS:
curl -sSL https://raw.githubusercontent.com/RightNow-AI/rightnow-cli/main/install.sh | bashWindows (PowerShell as Admin):
irm https://raw.githubusercontent.com/RightNow-AI/rightnow-cli/main/install.ps1 | iex
# Visit OpenRouter (no credit card needed!) https://openrouter.ai # Sign up with Google/GitHub for instant access # Copy your API key from the dashboard
rightnow
# Paste your API key when prompted (one-time setup)You: Create a vector addition CUDA kernel
RightNow: [Creates optimized kernel with detailed explanations]
- Learn CUDA: "Explain how CUDA threads work"
- Write Code: "Create a simple matrix multiplication kernel"
- Fix Errors: "Help! My kernel crashes with error X"
- Optimize: "Optimize this kernel for memory coalescing"
- Debug: "Find the race condition in my code"
- Analyze: "Profile this kernel and suggest improvements"
- No credit card required
- Free models included (Google Gemini, Meta Llama)
- Upgrade to premium models when needed
- General Assistant: Helps with any CUDA task
- Optimizer: Maximizes performance
- Debugger: Finds and fixes bugs
- Analyzer: Explains and improves code
- Read and write CUDA files
- Analyze performance bottlenecks
- Generate optimized variants
- Monitor GPU status
- Execute bash commands
# Start interactive mode rightnow # Inside RightNow: /models # List available AI models /gpu # Show GPU status /clear # Clear conversation /help # Show help /quit # Exit
Creating a Kernel:
You: Create a parallel reduction kernel
RightNow: [Writes complete kernel with shared memory optimization]
Optimizing Code:
You: Optimize my matrix multiplication for RTX 4090
RightNow: [Analyzes and provides optimized version with 10x speedup]
Debugging:
You: My kernel gives wrong results for large arrays
RightNow: [Identifies overflow issue and provides fix]
- Google Gemini 2.0 Flash - Fast and capable (default)
- Meta Llama 3.2 3B - Efficient for simple tasks
- GPT-4o - Most capable overall
- Claude 3.5 Sonnet - Best for complex code
- Gemini 1.5 Flash - Fast with huge context
Switch models anytime with /models command!
- Python 3.9+
- Any OS (Windows, Linux, macOS)
- Internet connection
- NVIDIA GPU (GTX 1650 or newer)
- CUDA Toolkit 11.0+
- 8GB+ RAM
# Add to PATH (Linux/macOS) export PATH="$HOME/.local/bin:$PATH" # Or reinstall pip uninstall rightnow-cli pip install rightnow-cli --user
# Get your free key at: https://openrouter.ai # Clear old key and re-enter: rm -rf ~/.rightnow-cli rightnow # Will prompt for new key
# Check NVIDIA driver nvidia-smi # Check CUDA nvcc --version # RightNow works without GPU (CPU simulation mode)
- Installation Guide: INSTALLATION.md
- GitHub: github.com/RightNow-AI/rightnow-cli
- Discord Community: discord.com/invite/sSJqgNnq6X
We welcome contributions! Please see CONTRIBUTING.md for guidelines.
Proprietary Non-Commercial License - FREE for personal and educational use. See LICENSE for details.
For commercial use, contact: jaber@rightnowai.co
- Discord: Join our community
- Twitter: @rightnowai_co
- Email: jaber@rightnowai.co
- Issues: GitHub Issues
Built with <3 by the RightNow AI Team
First GPU Native AI Code Editor