Bitcoin Vault Solutions Compared: Time-Locked Custody
Compare Bitcoin vault custody solutions using OP_VAULT, miniscript timelocks, and multisig recovery mechanisms. Covers clawback paths, key ceremonies, and tradeoffs.
Bitcoin Vault Custody Approaches Compared
A Bitcoin vault is a custody mechanism that enforces a mandatory delay before coins can be spent, with an emergency recovery path that allows the owner to claw back funds during the delay window if theft is detected. Unlike a standard multisig wallet, which protects against key compromise through threshold signing, a vault adds a temporal dimension: even if an attacker obtains the spending keys, the time delay gives the owner a window to intervene.
Several competing approaches to Bitcoin vaults exist today, ranging from protocol-level proposals that require a soft fork to wallet software that works with Bitcoin's existing script capabilities. The following table summarizes the four main approaches.
| Approach | Requires Soft Fork | Status | Clawback Mechanism | Key Deletion Required | Best For |
|---|---|---|---|---|---|
| BIP-345 / BIP-443 (OP_VAULT / CCV) | Yes | BIP-345 closed (May 2025); BIP-443 draft | Consensus-enforced recovery path | No | Ideal vault design (pending activation) |
| Liana (miniscript timelocks) | No | Production (v13+) | Recovery key unlocks after inactivity period | No | Inheritance, personal recovery |
| Revault (multisig vaults) | No | Proof of concept | N-of-N cancel transaction | No | Organizational treasury |
| Pre-signed transaction vaults | No | Prototype (Bryan Bishop, 2019) | Pre-signed clawback transaction | Yes (ephemeral keys) | Advanced users, research |
BIP-345 and BIP-443: Consensus-Level Vaults
BIP-345, authored by James O'Beirne, proposed two new opcodes: OP_VAULT and OP_VAULT_RECOVER. These worked alongside OP_CHECKTEMPLATEVERIFY (BIP-119) to enforce vault logic at the consensus layer. When coins were locked in a vault output, any withdrawal triggered a mandatory delay period. During that delay, the owner could broadcast a recovery transaction to sweep funds to a pre-committed recovery address.
In May 2025, O'Beirne withdrew BIP-345 in favor of BIP-443 (OP_CHECKCONTRACTVERIFY, or CCV), a more general-purpose covenant opcode. CCV can replicate all vault behavior that BIP-345 enabled, with additional capabilities: it can replace multiple tapleaves in a single operation and supports partial withdrawals where the remainder is automatically re-vaulted. O'Beirne stated that CCV has a smaller interface and implementation in the script interpreter than the vault-specific opcodes.
Neither BIP-345 nor BIP-443 has been activated on Bitcoin mainnet. BIP-443 remains a draft proposal with no activation timeline. For a deeper analysis of the covenant proposals that enable vault designs, see our research on Bitcoin covenants explained and BIP-345 OP_VAULT custody.
Liana: Miniscript Timelocks for Recovery
Liana, developed by Wizardsardine, is a Bitcoin wallet that uses miniscript to encode spending policies with time-locked recovery paths. The primary spending key is always available. A secondary recovery key becomes spendable only after the wallet has been inactive for a configurable period, up to approximately 15 months.
Liana uses relative timelocks (OP_CHECKSEQUENCEVERIFY) rather than absolute timelocks. This means the recovery window is measured from the last time the coins were moved, not from a fixed calendar date. The owner can reset the clock at any time by making a self-transfer before the timelock expires. Multiple recovery paths with different timelocks and key sets are supported: for example, a 3-month path for a trusted co-signer and a 12-month path for an inheritance key.
Liana does not strictly implement a vault in the consensus sense. There is no clawback transaction or watchtower-driven cancellation. Instead, the recovery key acts as a fallback that activates only when the primary key holder stops transacting. This makes Liana well suited for inheritance planning and dead-man-switch custody, but it does not protect against an attacker who compromises the primary key and spends immediately.
Note: Liana requires users to back up the wallet descriptor alongside their seed phrases. Without the descriptor, the miniscript spending policy cannot be reconstructed and funds may be unrecoverable.
Revault: Multisig Organizational Vaults
Revault is a multi-party vault architecture designed for organizations that need shared custody with spend-approval workflows. It uses N-of-N multisig combined with pre-signed transactions to enforce a two-phase spending process: stakeholders first co-sign an "unvault" transaction that begins a time-locked withdrawal, then a second transaction executes the actual spend after the delay expires.
During the delay window, any stakeholder can broadcast a cancel transaction to sweep funds back to the vault. Cosigning servers can enforce additional spending policies on the final transaction, though the Revault developers have noted this is suboptimal without covenant support: policies can only be applied to the unvault transaction, not to the downstream spend.
Revault remains a proof of concept. It does not protect against deliberate refusal to sign (a liveness failure), which the designers expect organizations to handle through legal agreements among stakeholders. The project shares developers with Liana: Antoine Poinsot (darosior) contributed to both.
Pre-Signed Transaction Vaults
The original vault concept, prototyped by Bryan Bishop in 2019, relies on pre-signed transactions with ephemeral key deletion. The process works as follows:
- Generate a temporary (ephemeral) private key
- Use it to pre-sign a complete set of vault transactions: the unvault, the timelocked withdrawal, and the clawback
- Securely delete the ephemeral key
- Once deleted, the only valid spending paths for those coins are the pre-signed transactions
If an unauthorized unvault transaction appears on-chain, a watchtower or the user can broadcast the pre-signed clawback transaction to recover funds before the timelock expires. The core security guarantee is that no one possesses the key to create new transactions, so the pre-signed paths are the only possible spending routes.
This approach has significant drawbacks. Withdrawal amounts, fee rates, and destination addresses must all be fixed when the transactions are signed. Any coins sent to a vault address after the ephemeral key has been deleted are permanently lost, since no key exists to sign new transactions spending them. Greg Sanders argued on Delving Bitcoin that none but the most sophisticated and well-funded operations can generate ephemeral keys in a way that is convincingly secure.
Covenant Approaches for Vault Design
A 2024 MIT Digital Currency Initiative report compared how different covenant proposals support vault constructions. The key tradeoffs are withdrawal flexibility, fee management, operational complexity, and on-chain cost.
| Covenant Proposal | BIP | Partial Withdrawals | Fee Flexibility | Vault Complexity | Non-Vault Uses |
|---|---|---|---|---|---|
| CTV (OP_CHECKTEMPLATEVERIFY) | 119 | No (outputs precomputed) | Can add fee-paying inputs | Low | Congestion control, payment pools |
| CCV (OP_CHECKCONTRACTVERIFY) | 443 | Yes (value splitting) | Flexible via tapleaf replacement | Medium | General covenants |
| APO/APOAS | 118 | No (outputs precomputed) | Inputs added more freely | Medium | Eltoo/LN-Symmetry |
| TXHASH (OP_TXHASH) | Draft | Yes (designer selects fields) | High (selective commitment) | High | Flexible introspection |
| OP_CAT | 347 | Possible (complex scripts) | Via script composition | Very high | General script composition |
| Pre-signed (no covenant) | N/A | No (fixed at signing) | None (fixed fee rate) | High (key deletion) | N/A |
The MIT DCI report concluded that CTV is suited to simple vaults with precomputed outputs, while CCV best supports vaults requiring partial withdrawals or trigger-time selection of a withdrawal address. None of these covenant proposals have been activated on mainnet. For a broader view, see the covenant proposals comparison tool.
Personal vs Institutional Vault Decision Framework
Choosing a vault approach depends on your security model, technical capabilities, and whether you are protecting personal savings or organizational funds.
| Criterion | Personal Use | Institutional Use |
|---|---|---|
| Primary threat | Key loss, death, incapacity | Insider theft, compromised signing device |
| Recommended today | Liana (miniscript recovery) | Multisig with organizational controls |
| Recommended (future) | CCV-based vault (pending soft fork) | CCV-based vault with cosigning servers |
| Key ceremony complexity | Low (1 primary + 1 recovery key) | High (N-of-N + policy servers) |
| Watchtower needed | No (Liana uses inactivity-based recovery) | Yes (for clawback during delay window) |
| Descriptor backup | Required (Liana) | Required (all miniscript-based setups) |
| Timelock management | Periodic self-transfer to reset clock | Automated via policy servers |
For personal users, Liana is the most practical vault-like solution available today. It works with existing hardware wallets that support miniscript (including BitBox02 and Blockstream Jade) and requires no soft fork. The tradeoff is that it protects primarily against key loss and incapacity, not against an attacker who steals your active signing key and spends immediately.
For institutions managing large Bitcoin treasuries, the current best practice is a multisig setup with geographic key distribution, as described in our multisig setup comparison. True vault functionality with clawback and time-delayed withdrawals will require a covenant soft fork. When that arrives, organizations will be able to combine multisig governance with consensus-enforced spending delays.
For users who want to hold both bitcoin and dollar-denominated stablecoins under self-custody, Spark provides a Bitcoin-native layer where BTC and stablecoins like USDB can be managed in a single self-custodial wallet without bridging to other chains.
Key Ceremony and Operational Complexity
Each vault approach places different demands on the key ceremony and ongoing operational maintenance.
Pre-signed transaction vaults have the highest ceremony complexity. The ephemeral key must be generated in a secure environment (ideally air-gapped), used to sign the complete transaction set, then provably destroyed. If destruction is incomplete, the vault's security guarantee collapses entirely. If the key is destroyed successfully but the pre-signed transactions are lost, the coins are permanently inaccessible.
Liana's ceremony is comparatively straightforward: generate a primary key and one or more recovery keys using standard HD wallet derivation, define the spending policy (including timelock durations), and back up both the seed phrases and the wallet descriptor. The main operational burden is remembering to make a self-transfer before the recovery timelock expires.
Revault requires a coordinated multi-party ceremony where all N stakeholders generate keys, exchange public key material, and pre-sign the initial vault and cancel transactions. This mirrors the complexity of setting up a corporate multisig, with the added step of deploying cosigning servers and watchtowers.
Covenant-based vaults (CCV, CTV) would significantly reduce ceremony complexity. Because the spending constraints are enforced by the network rather than by pre-signed transactions or destroyed keys, users would only need to set up their keys and define the vault policy in a tapscript tree. No ephemeral keys, no watchtowers for basic setups, and no descriptor backups beyond the standard wallet backup.
Frequently Asked Questions
Do I need a soft fork to build a Bitcoin vault today?
No. You can build vault-like custody today using existing Bitcoin script capabilities. Liana uses miniscript timelocks to create recovery paths that activate after a period of wallet inactivity. Pre-signed transaction vaults work with current script but require ephemeral key deletion, which introduces significant operational risk. True consensus-enforced vaults with clawback mechanisms will require a soft fork activating CTV (BIP-119), CCV (BIP-443), or a similar covenant opcode.
What is the difference between a vault and a multisig wallet?
A multisig wallet requires multiple keys to authorize a transaction, but once the threshold is met, the spend executes immediately. A vault adds a mandatory time delay before funds can be withdrawn, with a recovery path that allows the owner to cancel the withdrawal during the delay. Multisig prevents unauthorized spending through key distribution. Vaults prevent unauthorized spending through time, giving the owner a window to react to theft.
Can I use Liana for Bitcoin inheritance planning?
Yes. Liana's primary use case is inheritance and recovery. You set up a primary spending key that you control and a recovery key held by your heir. The recovery key only becomes spendable after a configurable period of inactivity (up to approximately 15 months). You must remember to make a periodic self-transfer to prevent the recovery key from activating prematurely, and your heir needs access to both the recovery seed phrase and the wallet descriptor to claim the funds.
What happened to OP_VAULT (BIP-345)?
BIP-345 was withdrawn by its author James O'Beirne in May 2025. He concluded that OP_CHECKCONTRACTVERIFY (BIP-443), a more general-purpose covenant opcode, could replicate all vault functionality with a simpler implementation. BIP-345 never activated on Bitcoin mainnet. BIP-443 remains a draft proposal with no activation timeline as of late 2026.
What are the risks of pre-signed transaction vaults?
Pre-signed vaults require generating and then securely destroying an ephemeral private key. If key destruction is incomplete, an attacker who recovers the key can bypass all vault protections. If the pre-signed transactions themselves are lost, the coins are permanently unspendable. Additionally, withdrawal amounts and fee rates are fixed at signing time and cannot be adjusted later. Any coins sent to a vault address after the ephemeral key is deleted are permanently lost because no key exists to sign new spending transactions.
How do vault timelocks differ from Lightning Network timelocks?
Both use Bitcoin's native timelock opcodes (OP_CHECKLOCKTIMEVERIFY and OP_CHECKSEQUENCEVERIFY), but for different purposes. Lightning uses timelocks to enforce channel state: if one party disappears, the other can claim funds after a delay. Vaults use timelocks to create a withdrawal delay that gives the owner time to detect and respond to theft. In Lightning, the timelock protects against an absent counterparty. In vaults, the timelock protects against a compromised key.
Which Bitcoin hardware wallets support miniscript for vault setups?
As of late 2026, BitBox02 and Blockstream Jade support miniscript and have been tested with Liana. Ledger added miniscript support for its Nano S Plus and Nano X devices. Coldcard and Foundation Passport have more limited or in-progress miniscript support. Because miniscript wallet setups require descriptor backups and policy-aware signing, not every miniscript-capable device integrates smoothly with every wallet coordinator. Check compatibility with your specific wallet software before committing to a hardware signer.
This tool is for informational purposes only and does not constitute financial advice. Vault designs involve complex security tradeoffs and potential risk of permanent fund loss if implemented incorrectly. Data is based on publicly available information as of late 2026. Covenant proposal statuses change over time. Always verify current BIP statuses and software versions before making custody decisions.
Build with Spark
Integrate bitcoin, Lightning, and stablecoins into your app with a few lines of code.
Read the docs →
