← All insights
RESEARCHZcash Forum · abdel

Zoro: a Cairo-based STARK light client

A draft design for Zoro, a light client implemented in Cairo that shows the validity of the Zcash chain with succinct STARK proofs, with no consensus change.

Zoro is a proposal for a zero-knowledge light client for Zcash implemented in Cairo. For a contiguous range of blocks it generates a succinct STARK receipt proving chain validity. Its status is a draft for technical review, and it is a standalone artifact that needs no consensus change. Without changing Zcash itself, it consists of an external prover and verifier and Cairo programs and tooling, and the receipt can be verified off-chain or, if TZEs (ZIP-222, ZIP-245) are adopted, on Zcash too.

The receipt proves three things. It checks that headers form a valid chain and satisfy the Equihash (n=200, k=9) proof-of-work check, that hashChainHistoryRoot, the FlyClient MMR commitment of ZIP-221, matches the headers, and that each block's transaction commitments are consistent with the ZIP-244 v5 digest rules. In the initial milestones it does not re-verify the Sapling and Orchard zero-knowledge proofs inside the STARK, but recomputes and checks the digests committed in the headers.

Uses it names are compressed SPV and cross-chain proofs that cover thousands of headers with a single receipt of tens to hundreds of KiB, mobile light clients that do not trust a server, and giving L2s such as Ztarknet a succinct statement of L1 state. It reuses the methodology and some Cairo components of Raito, which has already implemented a consensus client and a recursive STARK pipeline for Bitcoin. The prover is Stwo (Circle-STARK), long ranges are handled by recursion, and the verifier is written in Rust.

The work plan has milestones 0 to 5. They cover, in order, the header chain proof, the FlyClient MMR (ZIP-221), v5 transactions and block commitments (ZIP-244), public shielded invariants such as no duplicate nullifiers, and recursive aggregation and the API. The last is feasibility R&D on verifying Orchard/Halo2 receipts inside the STARK, which will report measurements and a recommendation on whether to proceed. Changes to shielded pool semantics or to the header or transaction formats are non-goals.

This summary is based on the original; check the original for exact details.