Stake DAO rewards original security vulnerabilities that cause users of in-scope production contracts to lose funds or rewards. Reports are assessed under the policy published when they are received.
Before you submit: read "Who Gets Paid" and the known issues on this page, then use the four-part evidence checklist below. Reports that show a loss under $10,000, or that lack this evidence, are closed after initial screening. Do not mass-submit AI-generated reports you have not checked: if we find this, we reject all your reports, including valid ones.
Who Gets Paid
A report is eligible only if it shows, on a fork of the deployed contracts at a recent block, that a user ends up with less than they are owed and cannot get it back themselves. Name the user (an address, a listing, a campaign or a gauge that exists on-chain at that block) and the amount.
The loss must be at least $10,000. It is measured like this:
- Counted at your fork block, on funds that exist in the affected contracts.
- Summed across all affected users in one realistic scenario.
- For a recurring leak, counted over one year at current rates.
Below $10,000, a report is not eligible. Stake DAO may still pay a thank-you for a real bug, at its discretion and with no entitlement.
Not eligible, however the code reads:
- A call that reverts and leaves funds where they are.
- A delay that ends when a keeper, a role or governance does its scheduled action.
- Funds left in a Stake DAO contract by the sender's own transaction, which governance can return.
- A condition that is false on-chain today and that only Stake DAO, governance or another protocol could make true. If you would call it "latent", do not submit it.
- A contract with no live user funds or campaigns: retired, shut down, dormant or being migrated.
- An issue described in an audit report, including acknowledged or unfixed findings, and the same finding on a sibling contract.
- Hardening: a missing zero check, dead code, or a comment that disagrees with the behaviour.
If your own Impact section says no funds are lost, do not submit.
Required Evidence
Use these four fields in your report:
- Target: chain ID, deployed address, implementation address for proxies or clones, and source matching the deployed code.
- Issue: affected functions, expected versus actual behavior, caller permissions and assumptions. Show that the required conditions exist or can arise through normal protocol use.
- Evidence: a minimal reproducible test or transaction simulation against deployed code, with dependencies and run instructions. For forks, include a recent block number and date; historical evidence must explain why the issue still applies. Include a control showing normal behavior without the reported condition.
- Impact: affected users and assets, with expected versus actual balances, entitlements or state. Separate measured harm from estimated totals. A protocol-wide valuation and proposed fix are optional.
Rules for the test:
- Paste the full test in the proof-of-concept field. "Available on request" is not evidence, and the report is closed.
- Run it on a fork of the deployed contracts. A test on mocks of Stake DAO contracts, on a local redeploy, or with code written over a production address is not evidence.
- The condition you rely on must exist at your fork block. Creating it by impersonating Stake DAO governance, a keeper or another protocol's owner invalidates the report.
- If the loss depends on an off-chain step, such as building a merkle root or a keeper run, show that step in the public off-chain code.
Disclose mocks, storage edits, injected balances and impersonated roles. Test setup cannot replace evidence that the required conditions and permissions are reachable in production.
Verify every claim before submitting, including claims produced by automated tools. Scanner output, code excerpts and passing test counts do not replace the evidence above.
If you claim funds are permanently lost or frozen, explain why existing claim, withdrawal, rescue, upgrade or governance paths cannot recover them. Otherwise, describe the recovery requirements, delay and any resulting harm.
AI-Assisted Reports
You may use AI and automated tools, but you are responsible for every report you send. Before submitting, check that the issue is real and that the contract is an in-scope Stake DAO contract.
Do not mass-submit reports you have not checked, such as reports on contracts that no longer exist, are no longer used or are not part of Stake DAO. If your submissions show this pattern, we reject all your reports without a reward, including reports that turn out to be valid.
We check your Mantis record. Several reports in a few days across bounties, on a new account, are handled under this rule.
Severity and Rewards
Stake DAO assesses severity from the loss you demonstrate, realistic preconditions, value at risk and recovery options. The severity you select when submitting, and any CVSS score, do not determine the reward.
| Severity | Demonstrated impact | Reward |
|---|
| Critical | Loss, unauthorized withdrawal or permanent freezing of user principal, of at least $10,000 | Up to 10% of the loss, capped at $100,000 |
| High | Theft or permanent loss of user yield or rewards, of at least $10,000 | Up to 10% of the loss, capped at $25,000 |
There is no Medium or Low tier. These amounts are ceilings, not guaranteed payments. Rewards reflect impact, originality and remediation value. Rewards are paid in USDC on Ethereum.
If an attacker cannot profit from the loss, the percentages above do not apply. Stake DAO reviews the report and decides the reward case by case.
Priority goes to the first eligible report with enough information to identify and reproduce the vulnerability. An incomplete or closed report that never established the issue does not pre-empt a later actionable report. Additional instances of the same root cause do not earn separate rewards.
Review
We aim to provide an initial response within 5 business days. Full assessment depends on complexity and volume.
Stake DAO decides whether to explain a decision. We give reasons only for reports we consider valuable; other reports may be rejected or closed without explanation.
Reports missing required evidence or falling outside scope may be closed without detailed technical review. Repeated submissions of the same claim without material new evidence may receive no further response.