The phone is the constraint.

One canonical argument now connects the phone, private discovery, spending, shared custody, recovery, and the limits of experimental Silent Payments. The earlier notes remain below as an archive.

Canonical paper · August 2026

From a Phone Wallet to Family Custody

Sofía's café-to-vault story, the complete mobile architecture, the threat model, verified signet evidence, explicit limitations, and the roadmap in one paper.

Read the canonical paper ›

Archived technical note · framing

A phone wallet

What a phone will not be (a full node, a thin client of someone else’s node), and what that forces: foreground-only, this-device keychain, Taproot-only, fresh and forward-only, one dependency.

Read mobile.md ›

Archived technical note · what’s mine?

The read side

Who scans, what they learn, what it costs. Compact filters by default, bounded mempool windows, an honest threat model, and why BIP37 is the historically important wrong answer.

Read read-side.md ›

Archived technical note · how it pays

The write side

Coin selection, a fee policy that admits it is blind, signing that holds the seed for one call, silent-payment send, and P2P relay that survives the app being killed.

Read write-side.md ›

Archived technical note · shared custody

Vaults

Two Taproot policies — NUMS + multi_a k-of-n, and MuSig2 n-of-n — coordinated by PSBTv2. No coordinator server.

Read vaults.md ›

Archived technical note · moving a wallet in

Import

There is no back-scan. The previous wallet ships its answers; this one checks them forward from a height. The bundle is the history.

Read import.md ›