Inspiration
Search tools find what is stored on your NAS. But some of the costliest problems in personal and small-business records are things that are not there: a skipped electricity bill, a rent receipt that never got filed, an insurance letter that never arrived. You can't search for a file that doesn't exist, so these gaps stay invisible until a penalty or a dispute shows up.
What it does
Filing Gap Detector is an AI NAS Skill that learns the rhythm of recurring documents already stored on your NAS (bills, receipts, statements, payslips) and reports which expected ones are missing, misfiled in another folder, or overdue. Every flag comes with its evidence: how many documents were seen, their dates, the usual arrival days, and the last one found.
The flow:
- Inventories the folder you name.
- Reads issuer, date and amount from documents, and from receipt photos.
- Shows the series it inferred so you can confirm or exclude them.
- Searches the folders you allow, so a misfiled bill isn't reported as missing.
- Delivers an evidence-first report. Saving it is optional and needs approval.
It is read-only until you choose to save, uses no scripts or external services, and keeps all data on the NAS.
How we built it
The Skill is a SKILL.md orchestration file plus reference files, following the UGOS Skill Development Standard. It delegates to the NAS file, document, photo and knowledge experts. Claude was used as an AI assistant to help draft the files and run the UGOS static checker.
Challenges we ran into
Unreadable scans must never be counted as missing, and a few false alarms would destroy trust. So the Skill requires at least four documents before judging a series, asks the user to confirm each series, and checks other folders before calling anything missing.
Accomplishments that we're proud of
The package passes the official UGOS static checker with zero errors and zero warnings. Every step is read-only except the optional, approval-gated save, and every flag carries its evidence.
What we learned
A Skill is instructions, not code, so detection has to be written as clear reasoning rules that delegate real work to NAS experts. Designing safeguards against false alarms mattered as much as detecting the gaps.
What's next for Filing Gap Detector: your NAS notices what's missing
The Skill has not yet been tested on a real UGREEN AI NAS. If selected as a finalist, the plan is to test real extraction and OCR output, scanned-PDF handling, date-inference accuracy and speed on large folders.
Built With
- ai
- markdown
- nas
- skills
- ugos
- ugreen
