Creative Toolbox

Inspiration

Before every project, content creators ask the same question: do I already have the footage I need? Photographers, YouTubers and video editors end up searching folders, copying huge files , end up losing many hours fetching their folders. Coming from enterprise backup and storage work environment, I also know that copies of important files should be checked but never trusted.

What it does

creative-toolbox is an Uliya Skill for UGREEN AI NAS. The creator names a shot list or project brief, the folders to search, and where the project should go. The Skill:

Reads the plan and shows the extracted shots for confirmation
Finds matching photos, videos, audio and images on the NAS
Checks real paths, sizes and duplicates
Presents a coverage report: each shot is found, needs review or *missing*and the total size
After the user confirms and approves, it creates a new project folder with one sub-folder per scene and copies the files and keeping the originals.
Verifies every copy with SHA-256
Saves the report in the project folder. It works with whole files only. It does not trim, convert or generate media. ## How we built it The Skill is a single SKILL.md following the UGOS Skill Development guidelines. It uses expert agent orchestration: the parent workflow only delegates to document-agent, photo-agent, file-agent and knowledge-agent, and each step passes real results to the next. I mapped every step to a documented capability in the creation kit (document extraction, photo search, file and media info, duplicate scan, folder creation and copy, sha256sum, and saving text), used the kit's static checker until it reported no errors, and designed explicit handling for completed, failed, cancelled, waiting-for-user and waiting-for-approval outcomes. Claude helped me draft ,revise and test the Skill . ## Challenges we ran into
Scope. My first idea was clipping segments out of footage. The supported capabilities don't include trimming, frame extraction or audio extraction, so I rebuilt the idea around what the NAS can simply do: find, check, organize and verify.
Honest limits. Whether the NAS tags video content and reads every shot-list format can only be confirmed on a real device. The Skill labels how each match was found and sends weak matches to "needs review".
No NAS in Round 1. Everything so far is validated statically, so I wrote down exactly what still needs testing: shot-list extraction, video matching, hashing speed on large files and the approval flow. ## What we learned Design from the capability list, not the wish list and to respect guidelines. Confirmations before every change, reports built only from real results, and verified copies make a tool people can trust with their footage. ## What's next for Creative Toolbox Testing it on a real UGREEN AI NAS if selected, tune matching, and add trimming-based features only if UGOS adds supported capabilities.