slothdesk: a local Gemma sorted my Downloads folder and found a secret I'd forgotten about
This is a submission for the Hacktoberfest Open-Source AI Challenge: Week 3 (theme: Slow and Steady). What I Built My Downloads folder had over forty files in it this week: old copies of my resume, a traffic challan named after its receipt number, statements, and a few files that really shouldn't be lying around in Downloads at all. I knew roughly what was in there. I just never wanted to open…
This week, a local developer named Kunal Siyag created a command-line tool called slothdesk to tidy up messy Downloads folders. The tool uses a local AI model called Gemma 4 to analyze each file in the Downloads folder and rename them with descriptive names. For example, a PDF file might become finance/2026-09-15-electricity-bill-october-2026.pdf.
The tool also moves files into sorted folders and generates a two-minute note that highlights the most important tasks, upcoming deadlines, files that shouldn't be in Downloads, and files that can be deleted.
Slothdesk ensures that no sensitive information is sent to the AI model. Files containing secrets, such as private keys or recovery codes, are identified and placed in a separate sorted/_lock-this-up/ folder. Private files, like bank statements or tax receipts, are identified but only the category (bank record) is shared with the note writer.
The remaining files are considered ordinary and assigned a new name with a one-line summary. All file names and contents are kept on the user's machine, and no secrets are opened by the model.
Kunal built the tool using Node.js, Ollama as the local AI model, pdftotext to extract text from PDFs, and ffmpeg to compress screenshots. The entire project is around 600 lines of code and takes about 7 seconds to analyze each screenshot. During testing, slothdesk identified a secret hidden in a screenshot of a recovery screen, which it correctly classified as a secret and moved to a separate folder.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.