Can I vibecode Zotero Storage?

Build a comprehensive, self-hosted, personal substitute for Zotero Storage, focusing on robust reference management, file syncing, and metadata organization.
Use exactly this modern stack: Next.js 15 App Router + TypeScript + Tailwind CSS for the frontend/API, and a local SQLite database (via Prisma or Drizzle) for lightning-fast data operations and easy portability.

Primary job: Construct a private research and reference workspace that seamlessly handles bibliographic data and file attachments (PDFs, images, web snapshots). The app must clearly distinguish between "Data Sync" (metadata, notes, tags, collections) and "File Sync" (the actual heavy attachments), mimicking the core architecture of reference managers without the SaaS storage quotas.

Core Features to Implement:
1. **Library & Collections:** Create a robust folder/collection structure allowing nested folders and tagging. Users should be able to organize items into multiple collections.
2. **Metadata Extraction & Management:** Implement a system to add references manually or import via URLs (fetching basic meta tags). Support various item types (Journal Article, Book, Web Page, Report). Include fields for Title, Authors, Abstract, Publication Date, DOI, and URL.
3. **File Attachment Storage:** Build a local file upload mechanism that securely stores PDFs and other documents on the server's filesystem, linked to their respective metadata records in SQLite. Ensure files can be easily downloaded or previewed in the browser.
4. **Rich Text Notes:** Include a markdown or rich-text editor for taking detailed notes attached to specific references, supporting code blocks, links, and basic formatting.
5. **Robust Search:** Implement full-text search across metadata fields (Title, Author, Abstract, Tags) and note contents.
6. **Storage Quota & Dashboard:** Include a simple dashboard showing the total size of stored files and the number of database records, helping the user monitor their self-hosted "quota."

Start from an empty folder and scaffold the complete working project.
Make the default mode single-user and entirely private. Store user data locally.
Do not add analytics, telemetry, or third-party accounts.
Ensure accessible keyboard navigation, focus states, and a clean, academic-focused UI.
Include clear empty states, loading indicators, and error boundaries.
Validate untrusted input strictly and ensure private file contents are served securely.

Write focused unit tests for the metadata parsing logic and database schema. Add one end-to-end smoke test using Playwright that proves the core loop: creating a reference, attaching a file, and finding it via search.
Create a detailed README with setup instructions, database schema explanation, backup procedures for the file storage directory, and current limitations.