Inspiration
Turn years of scattered family photos into a travel history you can revisit, privately on your NAS. The idea comes from the owner's own experience of manually reconstructing every trip, study-abroad stay and outing with AI help from 15,776 photos across 63 folders (2017-2026), and finding folder names that disagreed with actual capture dates.
What it does
Photo Trip Timeline makes that workflow repeatable as a UGREEN NAS Skill: it suggests events from an authorized library, shows the evidence, asks about contradictions, and saves an approved Markdown timeline with optional event albums. Photos never leave the NAS by design: verified local processing is required before private input, with no cloud fallback. You control the scope and approve every write.
How I built it
The Skill is written in Markdown for UGOS / Uliya (SKILL.md plus linked references) and uses the registered file-agent, photo-agent and knowledge-agent. No custom runtime code is included in the Skill ZIP. Local authoring and validation used OpenAI Codex, Python, ReportLab, pypdf, pypdfium2, Node.js, Git and uv.
What I learned
Folder names are not reliable evidence of when a trip happened. Capture dates often disagreed with them, so the Skill treats contradictions as questions for the user instead of guessing.
Challenges and current status
The Skill instructions, capability references and artifact utilities are built; official static checks and 11 local validator tests pass. On the finalist NAS, validation will cover installation, metadata access, local processing, timeline saving and exact album membership. Monthly review batches keep the process manageable; missing dates or GPS become questions, and a preview remains useful when write safeguards are unavailable.
AI disclosure
OpenAI Codex assisted official-source research, Skill and proposal authoring, packaging/PDF utilities, and local validation. The Skill and core workflow were newly written on October 26, 2026. Submission development accessed no private photos; the idea arose from the owner's earlier manual, AI-assisted review of their own library. Finalist NAS validation will establish runtime model selection and local execution before private input.
Log in or sign up for Devpost to join the conversation.