📦 StackStore

AI-Generated, OS-Level Development Environments. Built for the Microsoft AI Unlocked Hackathon.

---

📄 View the 6-Slide Presentation Deck (PDF)

🎥 Watch the 3-Minute Demo Video:

---

🛑 The Problem: Dependency Hell

Modern software development is plagued by environment configuration. Reviewing a simple GitHub repository often requires hours of installing specific Node versions, resolving Python virtual environment conflicts, and hunting down missing API keys.

Existing solutions are insufficient:

• Virtual Machines are too heavy, slow to boot, and expensive to host.
• Docker requires manual creation and maintenance of Dockerfile configurations.

💡 The Solution

StackStore completely eliminates local configuration. A developer simply pastes a repository URL into the StackStore client. Our AI Orchestrator utilizes the rapid reasoning capabilities of the Gemini LLM to analyze the codebase, infer the exact runtime requirements, catch missing secrets, and output a strict StackSpec JSON blueprint.

Our custom C++ Engine then parses this blueprint, utilizing native Linux kernel system calls to instantly carve out a zero-cost, localized, and securely isolated execution environment.

🏗️ Decoupled Microservice Architecture

StackStore is not a monolithic script; it is built as a robust, decoupled ecosystem:

1. Desktop UI (React/Electron): A sleek, cross-platform control panel for developers to input repositories, manage missing environment variables securely, and track environment states.
2. AI Orchestrator (Python): The intelligence layer. It reads the repository architecture and generates a standardized StackSpec blueprint defining the required OS state.
3. Core Engine (C++): A headless, high-performance binary. It reads the JSON blueprint and executes raw Linux clone() syscalls to isolate Process IDs (CLONE_NEWPID), Mount points (CLONE_NEWNS), and Hostnames (CLONE_NEWUTS), locking the code into a secure chroot-style sandbox.

🚀 Getting Started

1. Build the Virtualization Engine (Linux Required)

Because the engine interacts directly with the Linux Kernel to create isolated namespaces, it must be compiled and run on a native Linux system (like Ubuntu).

```bash cd engine make ```

2. Start the AI Orchestrator

```bash cd ai_brain python3 -m venv venv source venv/bin/activate # (Or .\venv\Scripts\activate on Windows) pip install -r requirements.txt python main.py ```

3. Launch the Desktop Client

```bash cd desktop_ui npm install npm run dev ```

🔮 Future Roadmap: Scaling the StackSpec

• Micro-sandboxes: Upgrading the C++ engine to parse a single StackSpec and split full-stack services into multiple isolated, networked namespaces without altering the core AI logic.
• Natural Language Environment Editing: Allowing developers to update and modify their active environment configurations instantly using simple voice commands.
• Windows/WSL Native Support: Expanding the C++ engine to hook into the Windows Subsystem for Linux (WSL2) API for native cross-platform execution.
• Hardware Passthrough: Enabling PCIe passthrough so sandboxed repositories can leverage the host machine's GPUs for local model training.