AutoDev Unit - the Artifact builder

[phodal]

Feature Proposal: AutoDev Unit – Reversible Artifact Builder / Skill

Description

AutoDev Unit aims to turn AI-generated outputs into reversible, executable artifacts, enabling a full lifecycle from generation → execution → re-editing.

Instead of stopping at source code download, AutoDev Unit packages AI-generated results into runnable desktop artifacts and allows users to load them back later to restore the original generation context and continue iteration.

This proposal focuses on two core use cases.

---

Use Case 1: Web Artifact (HTML / JS)

As a user,
I want to generate a small Web application (dashboard, tool UI, visualizer) with AI, export it as a standalone desktop app, and later load it back to continue editing with full context.

Expected Behavior

• AI generates a Web app (HTML / JS / CSS).

• User can:

  • Preview and interact with it.
  • Export it as a single executable file (.exe / .app).

• The exported app:

  • Runs without external dependencies.
  • Embeds its own source code and generation context.

• User can drag the executable back into AutoDev Unit:

  • The original chat, prompt, and code are restored.
  • User continues with: “Add a new chart”, “Change layout”, etc.

Problem It Solves

• Web artifacts today are one-way exports.
• Once downloaded, the AI context is lost.
• Packaging Web apps usually requires heavy tooling (Electron/Tauri).

---

Use Case 2: Python Script Artifact

As a user,
I want to generate a Python script with AI, run it locally without setting up Python or dependencies, and later reload it to continue evolving the script.

Expected Behavior

• AI generates a Python script (e.g. scraper, data processor, automation tool).

• Dependencies are declared inline (e.g. via PEP 723).

• User can:

  • Run it instantly.
  • Export it as a standalone executable.

• The executable:

  • Bootstraps its runtime automatically.
  • Requires no manual environment setup.

• Dragging the executable back into AutoDev Unit:

  • Restores source code + chat history.
  • Allows iterative changes like “Add CSV export” or “Optimize performance”.

Problem It Solves

• End users struggle with Python environments and dependency setup.
• Generated scripts lose context once downloaded.
• Existing packaging solutions are slow or heavyweight.

---

Core Idea

Executable artifacts should be reversible.

An exported artifact is not a dead binary, but a container that embeds:

• Source code
• Dependency metadata
• Original generation context

This enables a closed-loop workflow:
Generate → Run → Export → Reload → Iterate