JishAtlas Devlog #2

From Prototype to a Usable File Organizer

Since the first devlog, JishAtlas has grown from a basic AI prototype into a much more complete file organization tool.

What I Built

• Added project detection for common software projects such as Python, React, C++, and other project types.
• Improved the scanner with project roots, file information, content analysis, and passthrough directories.
• Added Local AI using a quantized Qwen model through llama.cpp.
• Added automatic Hugging Face model downloading and local model storage in %LOCALAPPDATA%\JishAtlas\models\.
• Improved the AI prompt and JSON output for more consistent organization.
• Added stronger handling of Windows paths and filesystem errors.
• Improved Organized Copy and Organize in Place modes.
• Expanded the Rich terminal interface with better previews, loading states, and summaries.
• Added local Gemini API-key configuration using .env.
• Started packaging JishAtlas into a standalone Windows executable with PyInstaller.

The Pipeline Now

Scan
  ↓
Project Detection
  ↓
Content Analysis
  ↓
Local AI / Gemini
  ↓
Organization Plan
  ↓
Validation
  ↓
Preview
  ↓
Apply

JishAtlas Devlog #3

Website, Packaging & Microsoft Store

JishAtlas has now moved from a development project toward a public release.

Website

Built and refined the official JishAtlas website with a professional product-focused design.

Added:

• Product introduction and feature sections
• Getting started guidance
• Documentation structure
• Interactive learning elements
• Direct EXE download option
• Microsoft Store download path for the upcoming Store release
• SmartScreen guidance for the standalone alpha release
• JishAtlas branding and product information
• Developer and GitHub project information

Windows Packaging

Packaged the PyInstaller build into an MSIX package using the Windows App Development CLI and Windows SDK tools.

The package includes:

• JishAtlas executable
• Windows application manifest
• Microsoft Store package identity
• Application assets and icons
• JishWorks publisher identity
• Full-trust desktop application configuration

The package was rebuilt after correcting the Microsoft Store package identity and publisher information.

Microsoft Store

Prepared JishAtlas for Microsoft Store distribution.

Completed:

• Created the JishAtlas Store product
• Reserved the product identity
• Configured publisher information
• Uploaded the MSIX package
• Completed pricing and availability settings
• Completed app properties
• Completed the IARC age-rating questionnaire
• Added Store listing information
• Added developer attribution
• Added restricted capability justification
• Prepared privacy policy information
• Submitted the application for certification

Release

The first public alpha is available as:

JishAtlas v0.1.0-alpha

The standalone Windows executable is distributed through the official GitHub release while the Microsoft Store submission is undergoing certification.

Next Steps

The next focus is completing the Microsoft Store release, improving the installation and update experience, gathering feedback from early users, and preparing future JishAtlas releases.

JishAtlas Devlog #1

Building the First AI Organization Engine

I started building JishAtlas with a simple goal: create a tool that can look at a messy folder and suggest a more logical way to organize it.

The first version focuses on getting the core pipeline working from end to end rather than trying to solve every edge case immediately.

First AI Pipeline

The initial AI organization pipeline is now working with Google Gemini.

JishAtlas can provide Gemini with information about the files and existing folders in a directory and receive an organization plan containing suggested folders, moves, and filenames.

This establishes the basic foundation of JishAtlas:

Folder
   ↓
Scanner
   ↓
File information
   ↓
Gemini
   ↓
Organization plan