This Skill groups a family's photos and videos from chosen folders into dated story albums and writes a short caption for each group, so the family can find and enjoy them again.

The user is a parent or grandparent whose family photos sit in scattered folders, unnamed and unsorted, so a birthday is hard to find and harder to show. The photos exist; when each was taken and who was there are lost.

The Skill reads the dates and places stored in each photo and video, groups items from one day or one event, and writes one short caption per group: date, place, and people named in the album data. It saves each group as a dated album to open, search, and browse.

The user says, 'Make albums from our summer photos,' and names the folders to use. The Skill shows the groups it found, with their caption and photo count, and waits for approval before creating any album. After approval it creates the albums and shows the first one to check.

Phone galleries sort by date but do not group a day into a story, write captions, or keep albums searchable by event. Cloud albums need uploads and hold the photos elsewhere. This Skill works on the family's NAS, groups by event and date, and writes captions the family can correct.

UGREEN AI NAS capabilities used: Albums finds photos by person, place, and time and creates the albums. Files lists the photos and videos in the folders the user names. Knowledge writes the captions and the album index as saved text the family can search.

Technical approach: Four steps — list, read dates and places, group and caption, then create albums after approval. Album, file, and knowledge agents handle grouping and creating, listing, and captions. Inputs are the source folders and a name pattern. The Skill needs read access to the photos and write access to the album folder, and creates nothing without approval. It uses the NAS local AI model, keeping everything on the NAS.

Current status: Designed, not yet implemented. The workflow, album rules, and caption format are written in SKILL.md. Nothing has run on a NAS; no NAS access was provided in Round 1. If selected, the Skill will be built with the UGOS Skill Creation Tool and tested on the assigned NAS.

Validation plan: Core test 1 — build albums from fifty photos across three events and check each photo lands in the right album with the right date. Core test 2 — search by event name and confirm the right album opens. Edge case — photos with no date stored, grouped by folder, with a note that the date is unknown.

Known limitations: Faces and places may be recognised wrongly, so a caption can name the wrong person or place; the family checks captions before approval. Videos with no date may be grouped by folder alone. The Skill depends on the NAS local AI model and on the Albums, Files, and Knowledge capabilities as documented.

Built With

  • local-ai
  • ugos
  • ugreen-ai-nas
Share this project:

Updates

Submission history