๐ฆฅ SlothDesk โ Let Your Desktop Stay Organized
Less sorting. More living. This is my submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady . What I Built I built SlothDesk , a local-first, AI-powered file organizer that helps take the repetitive work out of managing a cluttered folder. We all have that Downloads folder filled with assignments, invoices, notes, images, archives, and documents with confusingโฆ
Introducing SlothDesk, a local-first AI-powered file organizer designed to simplify the hassle of managing a chaotic folder full of documents with unclear names. This tool is perfect for those overwhelmed by the repetitive task of sorting through assignments, invoices, notes, images, archives, and confusingly named files. SlothDesk leverages a small open-weight language model that operates locally using Ollama, with Qwen2.5 1.5B being utilized for the categorization tasks.
What sets SlothDesk apart is its slow, careful, and reversible approach. It uses Ollama with Qwen2.5 1.5B to process file contents on the user's own computer. The software operates on a dry-run basis, unveiling the proposed organization of files before any actual movement takes place. Built in Python, the tool ensures safe file operations by avoiding overwriting existing files, validating paths, handling filename collisions, and maintaining a record of file movements for easy review and reversal.
The evaluation process of SlothDesk goes beyond guesswork, incorporating an evaluation harness to gauge the effectiveness of the chosen model in categorizing files. Lightweight in setup, the tool runs solely on a regular computer without requiring a GPU, making it accessible to a wider audience.
The project, available on GitHub at https://github.com/kvijay0611/slothdesk, includes the Python implementation, automated tests, configuration examples, evaluation code, and comprehensive documentation. Released under the MIT License, SlothDesk is open for feedback, bug reports, and contributions from the community.
The AI's performance was evaluated using a synthetic evaluation set. Qwen2.5 1.5B managed an 83% accuracy on the development set and 75% on the held-out set. While these results serve as an indication of the model's performance rather than a guarantee of accuracy on every real-world folder, they underscore the importance of the evaluation process.
In the demo run, SlothDesk successfully organized 11 files into categories such as Study, Finance, Work, Medical, Personal, Images, Identity, Installers, and Archives. The CLI estimated a time saving of approximately six minutes, based on an assumption of 30 seconds per file. Although this is an estimate, it highlights the repetitive work that SlothDesk is designed to minimize, with the bigger payoff being the ability to review an organization plan, apply it, inspect the changes, and undo if necessary.
Written by urgent.news from Dev.to's reporting โ not their text. Machine-written โ may contain errors; check the original before relying on it.