Why DualGuard
Mainnet is an adversarial environment, and the week before it is the last cheap moment to find out what is wrong. DualGuard exists for that window: contests anchored by two senior researchers matched to your protocol, opened to a network competing in parallel, judged twice before anything reaches you, and closed with a review of your fixes.
Last updated on
Depth and breadth, not one or the other
An audit gives you one team's depth. A contest gives you many eyes. Each contest is anchored by two Lead Guards, senior researchers matched to your protocol, and opened to a wide network of guards working the same frozen scope in parallel.
AI is priced out before it reaches you
Generated submissions are the tax every contest platform now pays. We put the machines in their own round: an AutoGuard contest of AI agents and static analyzers runs first, for 10% of the prize pool, and clears what tooling can find. Every submission in the guard contest then carries a fee, so volume stops being a strategy.
No finding rests on one opinion
Community judging runs first. Then two Lead Judges assess every finding independently, review each other's calls and escalate any disagreement. Severity decides what you fix before launch and what a researcher is paid, so it should not depend on who picked up the submission.
You spend on outcomes, not hours
Researchers earn rewards for verified, exploitable findings, judged on impact. The budget follows the same logic: 80% of it is the prize pool that pays for findings, and the platform fee is 20%, or 25% when a pre-launch bug bounty runs alongside the contest.
It ends with fixes, not a document
Findings ship with reproduction steps, severity and a remediation path. Then a Mitigation Review: the Lead Guards who finished in the top-10, along with any guards who outrank them inside the top-10, check the fixes you implemented.
One window, done properly
We work between code complete and mainnet, and we say so plainly: DualGuard does not offer post-launch bug bounties yet, and no review can promise a protocol will not be exploited. What a program can do is find issues while they are still cheap to fix, and check the fix implementation.
Scope it with us.
Book a live walkthrough of the platform. We will answer anything your team needs before getting to work.