Inspiration
Every family has the same problem: years of photos pile up on phones and get dumped onto a NAS "to organize later" - and later never comes. Cloud photo services solve this only if you hand them your entire family archive. We wanted the organization without the upload: an AI skill that lives where the photos already live.
What it does
Family Archivist scans a photo and video library on a UGREEN AI NAS, finds exact duplicates and blurry shots, groups media into events, writes captions and a story timeline for each event, and builds shareable albums. It is approval-first: it only ever proposes a cleanup plan, and nothing is moved or deleted until the owner approves it.
How we built it
The skill follows the Uliya Skill Development Standard. A SKILL.md defines the user goal, required inputs, CLI capabilities (scan, dedupe-report, cluster-events, caption-events, build-albums, apply-plan, undo), and an agent orchestration workflow: a planner agent sequences the steps while a verifier agent independently re-checks every proposed move or delete against the approved plan. Duplicate detection uses content hashing, already prototyped locally; captions will use the on-device vision model via UGOS Skills Hub. The project was designed and drafted together with an AI assistant (Muse), disclosed in SKILL.md as this hackathon requires.
Challenges we ran into
Round 1 provides no NAS to test on, so the workflow had to be verifiable headlessly: deterministic steps (hashing, clustering) are kept separate from model-dependent steps (captions), and destructive actions are isolated behind approval and verification so a bad plan can never touch the originals.
Accomplishments and what's next
Round 1 deliverables are complete: the proposal and the pre-test skill package. If selected as a finalist, we will run the skill on the assigned UGREEN AI NAS iDX6011 Pro, validate large-library performance and caption quality, and execute the test plan from the proposal: a 5,000-photo dedupe test, an event-clustering test, and a corrupted-file edge case.
What it does
How we built it
Challenges we ran into
Accomplishments that we're proud of
What we learned
What's next for Family Archivist
Built With
- ai-agent
- photo-management
- skills-hub
- ugos
- ugreen-ai-nas
