# Stake to Audit, explained

**After an audit, what is the auditor still responsible for? A plain account of the problem, a real example and what already exists.**

Explainer · 14 November 2026 · Plain-language companion to Initiative 3 of 3, TheDAO Security Fund Round 2. Research stage: nothing here is live.

Online: https://proofofaudits.com/explained/stake-to-audit

## In short

An audit report describes what was reviewed. It usually doesn't say what the auditor remains responsible for once the work is done, so when an audited protocol is hacked, the protocol and the auditor can disagree about whether the bug was ever in scope.

Stake to Audit tests a simple idea: an auditor may choose to put a limited bond behind a written scope, and independent reviewers decide any claim from the evidence. We want to find out whether protocols and auditors will use it.

## Live demos

- [Demo (recorded on a local Anvil chain)](https://proofofaudits.com/stake-to-audit/demo)
- [Bonded campaigns](https://proofofaudits.com/bonded-campaigns)

## 1. The problem

A protocol pays for an audit and launches. Months later it is exploited. The protocol says the auditor should have caught the bug. The auditor says that code, or that kind of issue, was outside the agreed scope. Often neither side can show the other a clear written answer.

Even when money is at stake, it doesn't settle the question. Someone still has to decide what was covered, using what evidence, and without a conflict of interest.

## 2. A real example

Euler lost about $197 million in March 2023. Euler's own account says the vulnerable code had been audited, and that the auditor also had money at stake through an insurance arrangement. The audit did not catch the missing health check ([Euler's account](https://euler.finance/blog/war-peace-behind-the-scenes-of-eulers-240m-exploit-recovery)).

The lesson we take is narrow: money at risk is not the same as a clear agreement about responsibility.

## 3. What already exists

*Existing approaches*

| Approach | What it does | What it leaves open |
| --- | --- | --- |
| [Sherlock Shield](https://docs.sherlock.xyz/coverage/sherlock-shield) | Financial coverage for code that went through a Sherlock audit. Payouts are decided by an onchain claims process, and Sherlock states it does not guarantee payment. | Tied to Sherlock's own audits; not a standard other firms can adopt. |
| [Sherlock bug bounties](https://sherlock.xyz/post/best-web3-bug-bounties-in-2026-the-highest-paying-programs-on-every-platform) | Researchers stake a small amount per report, refunded if the issue is valid. | Uses staking to filter reports, not to back a review. |
| Audit contests and audit firms | Pay for findings during the review period. | Nothing written down about responsibility afterwards. |

What is missing is an open, voluntary commitment that any auditor and any protocol can use: a written scope, a capped bond, a fixed period, and public claims and decisions.

## 4. What Stake to Audit would add

- A written agreement before the review: the exact code, what the auditor is responsible for, for how long, and the most they can lose.
- A bond the auditor chooses to put up, held in an escrow contract.
- Independent reviewers who decide any claim from the evidence. Nothing is paid out just because a hack happened.
- Every claim and decision published, including rejected ones.

Today this is research: eight contracts with the author's own Foundry tests, invariant tests and a Slither run, and seven scenarios recorded on a local Anvil chain with valueless test tokens and fictional auditors. The demo page shows every recorded transaction with the balances before and after it, and reads the test results from the report files ([demo](https://proofofaudits.com/stake-to-audit/demo)). Nothing is deployed, there are no live bonds, and the contracts have not had an independent security review.

## 5. What it does not claim

- It is not insurance, and it does not promise to cover user losses.
- A bonded review is not proof that code is safe.
- Nothing shows yet that bonding makes audits better. That is what the pilots are meant to test.

## References

- [Euler: behind the scenes of the 2023 exploit recovery](https://euler.finance/blog/war-peace-behind-the-scenes-of-eulers-240m-exploit-recovery)
- [Sherlock Shield documentation](https://docs.sherlock.xyz/coverage/sherlock-shield)
- [Stake to Audit demo (recorded on a local Anvil chain)](https://proofofaudits.com/stake-to-audit/demo)
- [Bonded campaigns](https://proofofaudits.com/bonded-campaigns)
- [Full whitepaper](https://proofofaudits.com/whitepapers/bonded-campaigns)
