Warranty Receipt Tracker

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Warranty Receipt Tracker

1. What does your Skill do? It builds a private, verified warranty register from the receipts, e-invoices, warranty cards and photographed till slips already stored on a UGREEN AI NAS, showing what is still covered, what expires soon, and the exact proof file for each item.
2. Who is your target user, and what is their problem? Households, families and small offices that buy appliances, phones and computers. When something breaks they cannot tell whether it is still under warranty, and the receipt is lost among thousands of photos, downloads and email PDFs. Warranties quietly expire unused.
3. What is your solution? One request ("what is still under warranty?") finds every receipt on the NAS, extracts product, seller, date, price and warranty term, calculates expiry and days left, and saves a register note. Optionally it tags the proof files.
4. What is the main user flow? The user names a folder or accepts the default directory, an alert window (30 days) and an optional default warranty rule. The Skill searches files and photos, extracts fields, and shows the register grouped as expiring, active, unknown and expired. After confirmation and platform approval, it saves the note and tags proofs.
5. How is your idea different from existing tools or common AI solutions? Warranty apps need manual entry or cloud uploads of private receipts. Generic chatbots read one file at a time and guess. This Skill uses data already on the NAS, combines file search with photo OCR to find receipts that file names hide, never invents missing terms (shows "unknown"), and links every row to its proof file.
6. Which UGREEN AI NAS features does your Skill use, and at which step? File search (step 1), photo OCR search and text recognition (step 2), document receipt and invoice extraction (step 3), knowledge save-text (step 5), file tags (step 6).
7. What is your technical approach? A Markdown orchestration Skill for the full main agent. It delegates read-only work to file-agent, photo-agent and document-agent, computes expiry in the conversation, then delegates approval-gated writes to knowledge-agent and file-agent. Each step passes only real upstream results. It handles completed, failed, cancelled, waiting_user and waiting_approval, handles empty results, and never overwrites files. Data stays on the NAS; no external service or credentials.
8. What have you built so far? What is not built yet? What still needs testing on a NAS? Built: SKILL.md plus references (field rules, register layout, test cases, disclosure); it passes the official checker with zero errors and warnings. Not built: tuning of prompts against real agent output. Needs NAS testing: OCR quality on crumpled slips, extraction accuracy across receipt formats, and the approval flow.
9. How will you test your Skill if you become a Finalist? Core 1: three PDF e-invoices, one expiring within 30 days; expect correct expiries and proof paths, nothing saved before approval. Core 2: a till-slip photo plus a warranty-card photo for the same model; expect both found and linked. Edge: a receipt with no warranty wording plus a corrupt PDF; expect status unknown, the PDF listed as unreadable, and a partial-result report.
10. What are the known limits, risks, or outside dependencies of your Skill? Accuracy depends on NAS OCR and extraction; faded slips may yield unknown fields. Ambiguous date formats are flagged rather than guessed. Warranty terms on manufacturer websites are not fetched. Third-party components: none. AI tools: an AI assistant (Grok Bot) helped draft the text; the entrant reviewed it.

Built With

• markdown
• ugreen-ai-nas
• uliya-skill-kit