Ideas from an external review (not implemented — product calls) #53

Collected while reviewing the three seQRets projects (PRs #43–#52). These are the items I did not implement because each is a product or format decision rather than a fix. Listed so they are not lost; take or close as you see fit.

1. Decrypt several files at once. Closed issue [Features] Support bulk file import for encryption #4 asked for bulk import. One password, N .ibitz files, N downloads — no ZIP, no new format. The drop zone already takes a FileList; today only [0] is used.
2. Argon2id as format v2. The one open audit item (SECURITY-AUDIT.md, Low Update README.md #1). Web Crypto has no Argon2, so it means a WASM implementation (~20 KB) inside the single file. A new version byte keeps every existing ciphertext opening; the recovery tool would need the same block. Best done together with 3.
3. Files above 100 MB via chunked AES-GCM (per-chunk nonce derived from a per-file random prefix + counter, each chunk authenticated, length-committed). Also format v2; same version-byte path.
4. Printable "emergency card." The QR of an encrypted text plus a one-paragraph instruction ("open ittybitz.app/ittybitz-recovery.html, paste, enter the password") on one page, for a safe or a notary. Reuses the existing QR; mostly a print stylesheet and a button.
5. Confirm the key file, too. Type the password twice before encrypting; an 8-word passphrase generator from the built-in BIP-39 list #52 adds "type the password twice"; the key file has the same failure mode (wrong file picked, discovered months later). A fingerprint of the key file's first bytes shown next to it — e.g. the first 8 hex chars of its SHA-256 — lets the user recognise it later without exposing it.

Not suggested on purpose: a BIP-39 seed generator (that is My-Seed-Phrase's job), cloud sync of anything, or a password manager integration (both contradict the "nothing leaves the page" posture).