Orphan Photo Albums — Track: Bring Memories to Life.
What does the Skill do?
Orphan Photo Albums finds photos in a NAS library that belong to no album, groups them into events by date, place, and people, and creates one album per event after the user approves it.
Target user
Families and hobby photographers with years of photos on a NAS. Their libraries grow into thousands of files, and pictures that never made it into an album become invisible, because everyone browses albums, not folders.
Solution
It scans the library against every existing album and rescues the leftovers. Orphan photos are clustered into events, and one album is proposed per event. Only approved albums are created; originals are never moved, renamed, or deleted.
Main user flow
The user asks for unorganized photos to be sorted. The Skill lists all albums, finds the photos that belong to none of them, and groups the orphans into events. It shows a proposal list. The user picks the albums they want, and the Skill creates them and reports what was made and skipped.
What makes it different
Photo apps organize by date or find duplicates. None ask the simple question behind this Skill: which photos live outside every album? That blind spot is the whole product. It also never moves or deletes anything, and approval is per album, not once for the whole run.
UGREEN AI NAS features used
Step 1 uses the photo library index to list albums and summarize the library by year and month. Step 2 uses photo search and metadata to find orphans and group them into events. Step 3 uses album creation, which asks the platform for approval before writing anything.
Technical approach
The Skill is a Markdown orchestration package built with the UGREEN Skill Developer Package v0.7.7 under the Uliya Skill Development Standard. It delegates all NAS work to the registered photo agent through the public ug-photo capability: album listing, timeline summary, photo search, and album creation. The parent Skill only coordinates, passing the real result of each step to the next, and handles failure and waiting explicitly. No scripts, executables, or credentials are included, and the on-device AI model does the grouping, so photos never leave the NAS.
Built so far
The full Skill package, SKILL.md and two reference files, passing the official UGREEN checker with zero errors and zero warnings. Grouping logic was also tested locally on synthetic photo data. Still needs NAS testing: a live run on a real device, album creation approval, and grouping quality.
Test plan as a finalist
Core case one — a library with a beach trip and a birthday among the orphans becomes two correct albums. Core case two — a library where every photo already sits in an album reports no orphans and creates nothing. Edge case — a four-day trip is split so no album spans more than three days.
Known limits
grouping depends on capture time, place, and people metadata, so photos without them may stay ungrouped.
Built With
- developer
- package
- skill
- ugos
- ugreen
