Comparison of Bitcoin Covenant Proposals for Vaults

[Lillian Wang]

Hello list,

I am sharing a report comparing simplified Bitcoin vault constructions using presigned transactions, CTV, APO/APOAS, TXHASH, CCV, and an OP_CAT-based Purrfect Vault construction. Comparison factors include partial withdrawals, withdrawal address commitment timing, flexibility for in-transaction fee management, operational complexity, and on-chain cost.

In the basic vault constructions we considered, CTV and APOAS have fully precomputed outputs and thus, like presigned transactions using SIGHASH_ALL, do not support arbitrary partial withdrawals. OP_CTV constrains the number of inputs, sequence values, and the current input index but not the input outpoints, allowing some flexibility in choosing fee-paying inputs. Meanwhile, APOAS omits more input information, meaning inputs can be added more freely.

TXHASH and CCV enable vaults with greater but differing functionality. TXHASH allows vault designers to select which transaction fields and input or output indices to commit to. This versatility provides more support for fee management and partial withdrawals at the cost of placing greater responsibility on the implementer to ensure the correctness and security of each vault. CCV can enforce how the current input’s amount is distributed across different outputs, supporting partial withdrawals that require remaining value to be revaulted. Additionally, CCV vaults can commit to the withdrawal address in the unvault transaction rather than only during vault setup. OP_CAT can support more arbitrary introspection and constraints, depending on the construction, but the construction considered in this report requires more complex scripts and witnesses and is thus less practical in its current form.

Here’s the full report:
https://raw.githubusercontent.com/Skyler-Cloud/Bitcoin-Vault-Comparison/main/bitcoin-vault-comparison.pdf

I have also opened a discussion on Delving Bitcoin:
https://delvingbitcoin.org/t/comparing-bitcoin-covenant-proposals-for-vaults/2877

Hoping it will be useful for other work consolidating analyses of vault constructions. I'd appreciate any feedback or corrections, especially with my interpretations of the APOAS and CCV constructions, comparison methodology, and other opcode combinations or vault properties.

Best,
Lillian Wang