Shared Photo Collector

Inspiration

After a trip, the memories are wonderful—but organizing the photos can be exhausting. Everyone uploads their pictures, and the shared folder quickly becomes a mix of group shots, food, scenery, screenshots, and receipts.

I wanted an easier way to find the photos that matter to each person. Instead of scrolling through hundreds of images, what if I could simply ask, “Collect the pictures of me, food, and scenery”?

That question inspired Shared Photo Collector.

What it does

Shared Photo Collector is an AI Skill designed for UGREEN AI NAS. It helps users turn a large collection of shared photos into a personal selection through conversation.

Users choose which authorized folders to search, then select categories such as People, Food and Drinks, Animals, Nature and Scenery, or Pictures of Me. For Pictures of Me, the user selects their existing indexed person entry on the NAS.

The Skill presents category counts, keeps uncertain matches available for review, and lets users choose what to collect. Photos can belong to several categories, but each selected image is included only once in the final collection.

It can organize the selection into an album or copy it into a separate NAS folder while preserving the originals. Users can then download the collection through the NAS interface.

How I built it

I used the Uliya Skill Kit and Codex to turn the idea into an orchestration Skill.

I first checked the published NAS capabilities to understand which parts of the workflow were supported. I then designed the Skill to coordinate the file expert for folder discovery and copying, and the photo expert for classification, indexed-person searches, and album creation.

I defined the categories, selection rules, approval boundaries, and behavior for missing information or incomplete results. I also created a conversational README to explain how someone would use it.

The Skill passed static validation with zero errors or warnings. Testing on a real NAS is still needed to verify the complete workflow and classification quality.

Challenges I ran into

The first challenge was finding a specific user problem worth solving. I explored several ideas before choosing the familiar frustration of sorting shared photos after a trip.

Another challenge was keeping the idea realistic. My original concept included identifying uploaders automatically and synchronizing photos directly to a device. After checking the available capabilities, I narrowed the workflow to user-selected contributor folders and collections prepared for manual download.

I also had to think carefully about overlapping categories. A picture of someone eating at the beach might belong to People, Food and Drinks, and Nature and Scenery. The selection rules needed to support that overlap without repeatedly collecting the same image.

Accomplishments that I'm proud of

I'm proud of turning an everyday frustration into a focused project with a clear audience and purpose.

Beyond coming up with the idea, I developed it into a structured Skill with category definitions, person selection, collection options, and handling for uncertain or incomplete results.

As someone new to both NAS and Skill development, creating a version that passed static checks was a meaningful first milestone.

What we learned

Before this competition, I knew very little about NAS and had never built a Skill. Researching UGREEN AI NAS helped me understand how shared storage and AI capabilities could work together to make stored content more useful.

I learned that building a good Skill involves more than describing an ideal result. It requires clear instructions, verified capabilities, sensible boundaries, and a plan for what happens when something goes wrong.

I also learned the value of narrowing a project. Checking feasibility early helped me shape an ambitious idea into a more practical workflow that I could explain, build, and test.

What's next for Shared Photo Collector

My next step is to test the Skill on a real UGREEN AI NAS using a varied collection of photos. I want to evaluate category accuracy, Pictures of Me results, and whether album creation and folder copying work as intended.

I also want to gather feedback on how easily users can choose categories and review uncertain matches, especially with larger collections.

Longer term, I would like to explore remembering previous collections and supporting easier device downloads or synchronization, if the platform exposes the necessary capabilities.

The goal is to make collecting shared memories feel as simple as asking for the moments you want to keep.

Built With

• codex
• uliya-skill-kit