Hello, I want to share a project I have been working on and would like feedback from the community.
The short version. I built a verifier for zero-knowledge STARK proofs that runs as an ordinary
Clarity contract. A complete proof, produced by a separate prover built on StarkWare’s open source Stwo library, is checked end to end by the contract and answers true or false. It needs no change to Stacks itself: no hard fork, no SIP, no new cryptography in the nodes. It works because STARK proofs are built entirely out of hashing and basic arithmetic, which are two things Clarity already has.
Why this might matter to you. Once a contract can check a proof, a class of applications
becomes possible on Stacks that is not possible today: compressing many transactions into one verified proof, proving that heavy off-chain computation was done correctly, trust-minimized bridges, and privacy designs with built-in accountability. The verifier is deliberately neutral infrastructure: open source, no token, not tied to any one application.
What is real today, stated carefully. On a simulated network running the real Stacks cost
rules, the demonstration proof verifies at about one percent of a single block’s compute budget. The verifier’s conventions were cross-checked value for value against StarkWare’s own proving code rather than assumed, and the whole thing is covered by more than 270 automated tests, including a large set of adversarial tests where every category of forged proof must be rejected. I also measured a cost model of the full-security version: it fits comfortably inside a single transaction, at roughly four to ten percent of a block.
What is not real yet, stated just as carefully. The working version runs at demonstration
strength, not production strength. Raising the security settings is the next engineering
milestone. The design has a written list of open questions that only an experienced STARK
cryptographer can settle, and none of this has been audited. I will not call it secure, and
nobody should build value on it, until that work is done. I have tried to be more honest about
the limits of this than is usual, because this field punishes overconfidence.
Links.
- Repository (contracts, tests, the Rust cross-check harness): [REPO LINK]
- Plain-language litepaper: [LINK]
- Technical litepaper: [LINK]
- The open questions awaiting expert review: [LINK]
What I am asking for.
- If you build on Stacks: would you use a shared proof-verification primitive, and what would you want its interface to look like? Even a one-line reply describing a use case helps.
- If you know the proof-systems world: the expert-review document is self-contained and I would be grateful for any portion of your attention on it.
- If you are involved with the working groups or the Foundation: I would welcome guidance on whether this belongs in the ecosystem’s shared toolbox, and what the right path looks like.
Happy to answer anything in the thread, including the skeptical questions.
Disclaimer
Project was guided using Claude Code Opus 4.8 & Fable 5 - using obra/superpowers & Trailbybits/skills audit agents. I am not an expert in cryptography or claim to be. Expert validation and review is needed.