Inspiration
Families and home-office users whose NAS became a dumping ground: backups, downloads, and scans pile into one folder as IMG_4021 and final_v2. Duplicates eat storage, nothing is findable, and cloud organizers mean uploading private family files.
What it does
nas-intake-organizer is a Skill for a UGREEN AI NAS that watches an intake folder, classifies each new file, renames it to a clean convention, files it into the right library folder, merges duplicates, and queues anything uncertain for human approval — all on-device, nothing leaves the NAS.
How it works
A Uliya orchestration Skill running entirely on the NAS. It scans intake, classifies by type and content, proposes clean names and destinations, detects duplicates by content hash, and applies only approved, reversible moves. Ambiguous files wait in an approval queue; every action is logged in plain language. Nothing is deleted silently.
UGREEN features used
Files (find, organize, deduplicate, tag), Documents (read, extract), Knowledge (save text for logs and queue summaries), Albums (filing photos into dated folders). All on-device.
How we built it
Built so far: the SKILL.md package (workflow, classification and naming rules, duplicate and approval policies) and a Python reference implementation tested locally, 25 of 25 checks pass. Not built: on-NAS execution. Needs NAS testing: capability calls, approval flow, and performance on large libraries.
Third-party components and AI disclosure
No third-party components inside the Skill package (Markdown only). The reference implementation uses only Python 3 standard library. AI tools: Muse (Anthropic) assisted with drafting, code, and test design; the author reviewed all output.
Built With
- markdown
- ugreen