Doomsday Tracker

Doomsday Tracker scans your documents for important expiry dates, organises them by urgency, and helps you avoid missing deadlines for things like passports, insurance, licences and warranties.
Doomsday Tracker is for anyone who keeps important documents digitally, whether it's individuals, families or small business owners. People often scan and store things like passports, insurance papers, rental agreements, licences and warranties, but rarely keep track of the expiry dates inside them. Those dates can be forgotten until it's too late, leading to unexpected expenses, expired coverage, missed renewals or even problems with travel and visas.
Doomsday Tracker searches a folder on my NAS, reads the documents using its built-in AI, and extracts details like the document type, owner's name and expiry date. It then groups everything by urgency: overdue, due within 30 days, within 90 days, or later, while setting aside documents with unclear dates for manual review rather than guessing. Once I review the results and approve them, it saves the summary as a note and tags the original files with their expiry year.
The user selects a folder on the NAS, lets Doomsday Tracker scan the documents, and reviews the organised expiry-date list alongside any files that need manual checking. They then approve the results, and the tool saves the summary note and tags the relevant files.
Most document organisers focus on keeping files organised by name, type or category, but Doomsday Tracker focuses on what needs attention and when. It connects every expiry date to its original file, flags anything unclear instead of making assumptions, and keeps everything on the NAS rather than sending sensitive documents to external services. I wanted to build something that doesn't just organise documents, but helps people avoid the consequences of forgetting about them.
It uses three UGREEN NAS agents: first, the file agent finds the documents; then the document agent reads expiry dates using local AI; finally, the knowledge agent saves the approved summary as a note and the file agent tags the files. Everything runs locally on the NAS, keeping users' documents private.
It is built as a single Skill written in plain Markdown, without code scripts. Each step passes its results to the relevant NAS agent, using actual outputs to guide the next step. It takes a folder and time window as inputs, handles different agent states, and requires user approval before making any changes.
The full Markdown workflow, agent-state handling, review list and disclosures are built and passed a clean check. Nothing remains unbuilt; real NAS testing is still needed for scan accuracy, approval prompts and file tagging.
I'll test with five sample documents to confirm the dates are extracted and grouped correctly. Next, I'll approve the results and verify the note is saved and files tagged. Finally, I'll test a blurry scan to ensure it's flagged for review without receiving a tag.
Accuracy depends on scan quality; runs on demand, requires NAS permissions, and has no external dependencies.