The Intelligent Development Environment Architect for Arch Linux
Key Features • How it Works • Installation • Getting Started • Tech Stack • Roadmap
ai-pkg is not just a package manager wrapper; it's a bridge between your intent and a fully functional development environment. By leveraging advanced LLMs (Gemini, OpenAI, Groq, and Ollama), it understands complex requests in plain English and transforms them into a curated, safe, and efficient installation plan.
"Set up a high-performance React frontend with Tailwind and a Python/FastAPI backend."
With one command, ai-pkg identifies the necessary Arch/AUR packages, scripts the environment setup, and provides a sleek TUI to review every single step before execution.
| Launch & Chat | Plan & Execute |
|---|---|
![]() |
![]() |
Built with Textual, the interface is designed for power users. It features a responsive layout, Catppuccin-inspired aesthetics, and smooth keyboard-driven navigation.
Swap backends on the fly. Whether you prefer the reasoning of GPT-4o, the speed of Gemini 2.5 Flash, the low-latency of Groq, or the privacy of Local Ollama, ai-pkg supports them all.
The built-in AI Filtering System is unique: if you deselect a package from the proposed plan, the AI automatically scans the remaining shell commands and strips away any logic that depended on the removed package. No more broken environment scripts!
graph TD
A[User Prompt] --> B{AI Engine}
B -->|Analyze| C[Package Identification]
B -->|Scripting| D[Shell Command Generation]
C --> E[Interactive TUI Plan]
D --> E
E -->|User Selection| F[Semantic Filter]
F -->|Validation| G[Safe Execution]
G --> H[Ready-to-Code Env]
Install the binary package via your preferred AUR helper:
yay -S ai-pkg-bin
# OR
paru -S ai-pkg-bingit clone https://github.com/rohankrsingh/ai-pkg.git
cd ai-pkg
./install.shLaunch the interface by simply running:
ai-pkg| Key | Action |
|---|---|
F2 |
Settings: Configure LLM providers and API keys |
↑ / ↓ |
History: Navigate prompt history in the chat |
Space |
Toggle: Select/Deselect packages in the plan |
r |
Run: Execute the finalized plan |
Esc |
Cancel: Return to chat or exit current view |
Ctrl+C |
Quit: Safe exit |
The application is fully configurable via the TUI (F2) or by editing the config file directly:
Location: ~/.config/ai-pkg/config.toml
[ai]
model = "gemini" # choices: gemini, openai, groq, ollama
gemini_api_key = "..."
gemini_model = "gemini-2.5-flash"
groq_api_key = "..."
# ... other provider settings- Language: Python 3.9+
- TUI Framework: Textual
- Styling: Rich
- HTTP Client: HTTPX
- Serialization: TOMLi
- Docker Integration: Automatically generate
Dockerfileanddocker-compose.ymlbased on the plan. - Community Presets: Share and download community-curated environment templates.
- Plugin System: Support custom post-install scripts and hooks.
- GUI Wrapper: A lightweight Electron/Desktop wrapper for non-terminal enthusiasts.
Stay current with the latest features:
yay -Syu ai-pkg-binTo remove the application and its local environments:
./uninstall.sh📜 Previous CLI Version Documentation (v1)
- 🧠 AI-Powered Intelligence: Understands context and recommends the perfect package combinations
- 🎯 Smart Package Selection: No more manual package hunting - AI picks the right tools for your needs
- 🛠️ Instant Environment Setup: Turn ideas into ready-to-code environments in seconds
- 🏗️ Project Bootstrapping: Creates and configures new projects with industry best practices
- ⚡ AUR Superpowers: Seamless integration with both official repos and AUR
- 🤝 Flexible Installation: Works with both
yayandparu- your choice! - 🧪 Safe Preview Mode: See exactly what will happen before making any changes
- 🎨 Framework Agnostic: Works with any tech stack - React, Vue, Django, Express, you name it!
Install directly from the AUR using your preferred helper:
yay -S ai-pkg-bin
# or
paru -S ai-pkg-bingit clone https://github.com/rohankrsingh/ai-pkg.git
cd ai-pkg
./install.shSet up your Gemini API key (required for v1):
export GEMINI_API_KEY=your_api_key_hereFor permanent setup, add the above line to your ~/.bashrc or ~/.zshrc.
Experience the legacy CLI-driven setup:
# Create a new React project with modern stack
ai-pkg "set up a react js app as test-app with tailwind, zod, react hook form"
# Set up a data science environment
ai-pkg "create a python data science environment with jupyter, pandas, and scikit-learn"
# Set up a development environment for backend
ai-pkg "set up nodejs backend environment with express, typescript and mongodb"
# Preview mode (shows what would be installed)
ai-pkg "set up a vue.js development environment" --dry-runThe AI will:
- 📋 Analyze your requirements
- 🎯 Select the appropriate packages
- ⚙️ Configure the environment
- 🚀 Set up the project structure (if requested)
- 📝 Provide any necessary post-installation instructions
| Option | Description |
|---|---|
--dry-run |
Preview commands without executing them |
--yes |
Auto-confirm all prompts |
--run-env |
Run environment setup steps automatically |
--aur-helper |
Choose AUR helper (yay or paru) |
Environment variables:
AI_PKG_AUR_HELPER: Set default AUR helper (yay/paru)GEMINI_API_KEY: Your Gemini API key
Distributed under the MIT License. See LICENSE for more information.

