Shelf: sorting Downloads offline with a 23 MB open model

Description and use

Shelf automatically sorts a Downloads folder offline with a 23 MB sentence‑embedding model, providing safety via dry runs and an undo log.

Problem

Users struggle to organize cluttered Downloads folders, especially plain‑text files whose content does not indicate a target folder.

Outcome

Shelf automatically organizes Downloads offline, using a 23 MB sentence‑embedding model to classify plain‑text files, while ensuring safety via dry runs, hard‑link moves, and an undo log.

Workflow

• 1. Run shelf on the Downloads folder to generate a dry‑run plan. 2. Review the plan and the model’s decision rationale. 3. Execute shelf --apply to move files. 4. Use shelf undo <folder> to revert the last run if needed.

Evidence

Shelf is a command-line tool that tidies a messy folder. Point it at your Downloads folder and it shows you where every top-level file would go.

Source

The file’s text is embedded once, and each folder is scored by its best-matching prototype. The top folder wins only if it beats the runner-up by at least 0.05 cosine similarity.

Source

Dry run is the default. Nothing moves without --apply.

Source

On ten synthetic text files that I wrote for this test, the real model got 7 right. Two went to Unsorted and one was wrong.

Source