JishAtlas Devlog

AI Organization Engine

Completed the core AI pipeline using Google Gemini. JishAtlas now scans folders, analyzes file contents and the existing folder structure, and generates a cleaner, more logical organization with improved folder hierarchies and file names.

Folder Scanner

Built a recursive scanner that collects files, folder information, and text file contents while safely skipping binary files. Added support for passthrough directories such as node_modules, .venv, .idea, and other generated folders so they remain unchanged during organization.

Preview and Apply

Implemented a preview step that shows every proposed rename and move before any changes are made. Users can either create an organized copy of the folder or apply the changes directly to the original.

Reliability Improvements

Refined the AI prompt to produce more consistent results, simplified the JSON response format, and fixed several path resolution and Windows filesystem issues encountered during testing.

Command-Line Interface

Added a cleaner command-line interface with a custom startup banner, improved output formatting, and organization summaries to make the overall experience more polished.

Next Steps

The next focus is project detection so software projects are preserved as complete units during organization. After that, I’ll be working on validation checks, undo support, and further improvements for organizing large real-world directories safely.