The 4-Layer Self-Custody Test: A Framework for Evaluating Wallets
Most wallet comparisons miss the point. Use this 4-layer test — keys, signing, recovery, exposure — to honestly evaluate any wallet's self-custody claim.
The phrase 'self-custodial' is used loosely. A genuinely self-custodial wallet must satisfy four independent properties — and most wallets fail at least one of them. The 4-Layer Self-Custody Test is a framework for cutting through marketing claims and evaluating any wallet's actual control profile. Apply it consistently and the answer to 'is this product really self-custody?' becomes unambiguous.
Layer 1 — Key Custody
- Question: who holds the private key?
- Pass: the user's device, hardware wallet, or a multi-sig where the user controls the quorum
- Soft pass: MPC where the user holds the threshold-required share — strict scrutiny on the recovery service
- Fail: a remote service holds any signing share by default — even if 'recoverable by the user'
- Test: can the user export a key or seed they can use independently of the vendor's app?
Layer 2 — Signing Authority
- Question: who signs the transaction?
- Pass: the user's local device signs, with no remote co-signer required
- Soft pass: a co-signer is required only above a user-defined threshold
- Fail: the vendor signs by default and the user 'requests' transactions through a UI
- Test: can the user sign a transaction with the vendor's servers offline?
Layer 3 — Recovery Sovereignty
- Question: can the user recover their wallet without the vendor's cooperation?
- Pass: a written seed phrase or hardware backup is sufficient — vendor irrelevant to recovery
- Soft pass: social recovery via guardians the user chose, with no vendor in the quorum by default
- Fail: vendor must approve recovery, or 'recovery' depends on a vendor-side database
- Test: imagine the vendor goes bankrupt overnight. Does the user retain access?
Layer 4 — Permission Exposure
- Question: what standing authority has the wallet granted to third-party contracts?
- Pass: bounded approvals (Permit2 with amount and expiry) and explicit standing-approval audit
- Soft pass: infinite approvals with a one-tap revoke surface that the user can audit at any time
- Fail: standing approvals are invisible to the user, or revoke surface does not exist
- Test: can the user list every contract that can drain their tokens, sorted by approved amount?
How Steyble Scores
Steyble passes Layer 1 (keys generated and stored on the user's device, hardware-wallet-compatible), Layer 2 (local signing, optional account-abstraction co-signers under user control), Layer 3 (12-word seed plus optional social recovery the user configures), and Layer 4 (Permit2 bounded approvals and a built-in approval audit dashboard). Apply this test to any wallet you are considering — most marketing-heavy products fail Layer 3 or Layer 4 within minutes of investigation.