Inspiration
Almost everyone has thousands of photos, and almost nobody knows how they actually shoot. Which focal length do I reach for most? Did it change this year? Is that expensive 85mm lens just gathering dust?
The idea started with one simple question: "What's my go-to mm, and how has it changed from March to August?" Every photo already records the answer in its hidden camera data (EXIF), but nobody ever asks. A UGREEN AI NAS felt like the perfect place to ask, because your whole photo history already lives there, privately, on your own device.
What it does
Pixora is a Skill for UGREEN AI NAS that turns your photos' camera data into a personal shooting story, then coaches you to grow.
Ask "What's my go-to mm since March?" and Pixora shows:
- Your signature focal length, the one you use most
- Your drift, month by month (for example, from 35mm to 85mm in six months)
- Your comfort zone, and how it widens or narrows over time
- Your lens loyalty, including lenses you've quietly stopped using
Then comes the fun part. Pixora suggests a practice challenge, like "40 frames with your forgotten 85mm in 30 days", saves it as a note on your NAS, and checks your progress whenever you ask, with an optional review album. Optional extras cover settings by subject, low-light habits, video clips, and a camera-shake risk flag.
Pixora reports habits, never photo quality, and nothing on your NAS changes without your approval.
How I built it
Pixora is a pure Skill, a SKILL.md that follows the Uliya Skill Development Standard, with no scripts. It directs three of the NAS's built-in experts:
- photo-agent reads the statistics, counts photos, and builds albums
- knowledge-agent saves reports and challenge notes
- file-agent finds and reads the saved challenge
I began by reading the UGOS Skill Creation Tool and its capability catalog, so every step relies on something the NAS can really do. The Skill defines its own metrics, such as the Comfort Zone Index
I checked the finished Skill with the kit's official checker, which passes with no errors and no warnings. I used Claude (an AI assistant) to help draft and refine the Skill text and this write-up, and I reviewed and edited everything myself.
Challenges we ran into
- Building only what the NAS can really do. Several early ideas, like detecting blurry photos or finding a moment inside a video, aren't supported, so I dropped them.
- No NAS yet for Round 1. I can't prove the counting works. I designed Pixora to be honest about that, so every report says how many photos it analyzed out of how many matched.
- Messy real-world data. Photos shared through messaging apps lose their camera data, camera clocks can be wrong, and phones report tiny focal lengths. So Pixora analyzes each camera separately and flags months with too little data.
- Resisting the urge to claim "quality." The NAS can't judge sharpness, so Pixora sticks to habits.
- Keeping it focused. A 500-word proposal forced me to cut features until one clear idea was left.
Accomplishments that we're proud of
- A complete Skill that passes the official checker cleanly
- An idea that ends in action (a tracked challenge and a review album), not just another chart
- A Skill built to never invent data: it reports coverage, marks thin months, and waits for approval before any change
What we learned
- A Skill is a brief, not a program. Clear instructions matter more than clever code.
- Read the platform's capability list first. The best ideas come from what the NAS can actually do.
- Honest limits make an idea stronger, not weaker.
- Hidden EXIF data tells a surprisingly human story about how a person sees the world.
What's next for Pixora
If I'm selected as a Finalist, the first job is testing on a real UGREEN AI NAS: whether the focal length and ISO filters accept ranges, how fast counting runs on large libraries, and how RAW files and video details behave. After that I'd like to refine the coaching loop based on real photographers' feedback and share Pixora through the UGREEN Skills Hub.
Built With
- claude
- markdown
- uliyaskillkit
