Can I vibecode Canary Mail?

Build a personal replacement for Canary Mail in an empty repository.
Use Tauri 2, React, TypeScript, SQLite, IMAP, SMTP, and an AI provider SDK (e.g., OpenAI/Anthropic); do not offer alternative stacks.
The core loop is: connect an existing mailbox, index messages locally, provide a fast, privacy-focused inbox with AI summarization, and send through SMTP.
Make the first run work locally with one documented command. Store all user data locally by default.

Data Model:
- `Accounts`: id, email, protocol_settings
- `Folders`: id, account_id, name, type (inbox, sent, archive)
- `Emails`: id, account_id, folder_id, thread_id, sender, recipients, subject, body_html, body_text, date, read_status
- `AiSummaries`: email_id, summary_text, generated_at
- `Drafts`: id, account_id, to, subject, body, tone_preference

User Journey & Screens:
1. Unified Inbox: A sleek, distraction-free list of emails. Include a "Smart Search" bar using natural language to filter local SQLite records.
2. Email Reader: Renders HTML securely in a sandbox. Includes a prominent "Summarize with AI" button that condenses long threads.
3. AI Compose Screen: Standard composer plus an "AI Copilot" sidebar to adjust tone (professional, friendly, concise) or generate drafts based on a prompt.

Technical specifics:
- Store credentials only in the OS keychain using Tauri plugins. Local emails stay in SQLite.
- Implement an IMAP sync engine to index headers and message bodies locally with resumable sync.
- Render HTML email in a locked-down sandbox and block remote images by default to protect privacy.
- Deliberately leave out: End-to-End PGP encryption, mobile push services, provider-scale spam filtering, and hosted read receipts.
- Write focused tests for the local IMAP syncing and the AI summary prompt generation logic.
- Create a README with setup, architecture, local data location, and how to configure the AI API key.
- Run the tests and build before finishing, then fix errors rather than merely describing them.