The Verifiable Launch: A Trust-Minimized Protocol for Audit-gated Token Deployment

Abstract

Token projects face a financing deadlock: a professional security audit is the single strongest pre-launch signal of contract safety, but audits cost tens of thousands of dollars and must be paid for before a token exists to sell, while contributors who could fund the audit have no way to verify what they are funding. Existing launchpads resolve this deadlock by inserting trusted parties (the platform, a review council, the founder's promises), which relocates the trust problem rather than removing it. We formalize the verifiable launch: a deposit-and-deployment protocol in which an escrow contract accepts exactly four kinds of input (money, clock time, cryptographic proofs, and per-user choices affecting only that user) and in which no actor ever attests, approves, or decides anything on behalf of another actor. A founder commits on-chain to a fixed-schema claims manifest describing the token's properties, together with a zkVM receipt proving that a complete implementation satisfying those claims existed before any money moved, without revealing the code. Contributions sit in a non-custodial escrow; on success, an audit fee derived deterministically from the proven code size is escrowed for a pre-named auditor; the audit report is bound on-chain either by the auditor's own typed-data signature or by a permissionless web proof of the report on the auditor's domain; machine-evaluated launch conditions over the report's finding counts gate deployment; a second receipt binds the post-remediation code to the same manifest; and the escrow itself deploys the token by CREATE2 and doubles as the distribution ledger. Every failure path degrades automatically to pull-based refunds. We give the mechanism as a formal state machine with exact verification predicates, analyze its security against adversarial founders, contributors, auditors, platforms, and proof-layer attacks, treat the incentives game-theoretically (including why paying the auditor independent of findings dominates outcome-contingent payment), and cost the hard engineering: compiling Solidity inside a zkVM, for which we define a tiered deployment path that upgrades from attested to fully proven verification by verifier substitution alone. We are precise about what the mechanism cannot do: it proves process, never quality; it removes discretionary trust, never market or legal risk; and a clean audit of an honest manifest does not make a good project. Section 8 records the seven design revisions the analysis forced on earlier iterations of the protocol.