← back to blog

When A Protocol You Deposited Into Gets Exploited: The First Hours

The first message is usually a rumour

It usually starts sideways. A post in a group chat says a protocol is “getting drained”. Someone shares a screenshot of an explorer page with large outflows. A security alert account posts that it’s seeing suspicious transactions on a contract. Within minutes, dozens of people are asking whether it’s real, and dozens more are answering with confidence they don’t have.

If you have funds deposited in that protocol, the next hour matters. Some actions in that hour can protect what’s left. Others make things worse, and a few hand whatever remains to a second attacker who’s waiting for exactly this moment of panic.

This piece is a runbook for that hour and the ones after it. It isn’t about your own mistakes, like signing a drainer approval, which is covered in a separate piece on recovery. This is about the protocol itself being hit: a bug in its contracts, a manipulated price feed, a stolen admin key, a hijacked website or a failure in something it depends on. You didn’t do anything wrong. Your funds are exposed anyway.

The honest starting point is that, in many serious exploits, there’s nothing a depositor can do in time. Attackers usually move faster than any human reading the news. But not every incident is like that. Some are slow, partial or limited to one part of a protocol. Some are frontend attacks where the contracts are fine and the danger is the website. Knowing which kind you’re in decides whether you should act, and how.

The order below is deliberate: confirm, classify, check your exposure, act only where it helps, avoid the second wave, and write it all down. It’s written for farmers, who often have deposits spread across many wallets and many protocols, and who need a calm process more than a fast one.

Confirm before you do anything

The first job is to find out whether the incident is real, and what the protocol itself is saying. Rumours are often wrong in the details, and sometimes wrong entirely.

Go to the protocol’s official channels directly, not through links in the thread where you heard the news. That means its verified social account, the announcements channel in its official Discord or Telegram, and its status page if it has one. Type the addresses or use bookmarks you saved earlier.

Look for a statement from the team. In real incidents, teams often post quickly, even if only to say they’re investigating and to tell users what not to do. The most useful early statements name which contracts or products are affected and whether users should withdraw, stop depositing or revoke approvals.

Cross-check. Official accounts can themselves be compromised, and a fake “urgent: migrate your funds here” post from a hijacked account is a classic follow-up to a real incident. If the protocol’s X account says one thing and its Discord announcements say another, or if a post pushes you to a new site you’ve never seen, treat it as suspect until confirmed elsewhere.

Independent security researchers and monitoring accounts are often the first to spot exploits, and their posts can help you understand what’s happening. Treat them as useful context, not instructions. They don’t control the protocol, and they sometimes get details wrong in the first hour.

On-chain evidence is the most reliable signal. If you know the contract addresses you’re exposed to, you can look at them on an explorer and see whether large, unusual outflows are happening. You don’t need to understand every transaction. A vault that normally sees small deposits and withdrawals suddenly sending most of its balance to one new address tells you something is wrong.

Work out what kind of incident it is

Not every “exploit” is the same, and the right response depends heavily on what’s actually broken.

A contract exploit means there’s a bug in the protocol’s smart contracts that an attacker is using to take funds. This is the case people usually imagine. If the affected contract is the one holding your deposit, the attacker may be draining it faster than you can react. If it’s a different contract in the same protocol, your deposit may be untouched.

An oracle or price manipulation means the contracts are behaving as written, but they’re acting on bad prices. An attacker might push a price feed or a pool ratio to borrow more than they should, or to liquidate others. Depending on the design, the damage can be limited to certain markets or collateral types.

An admin key compromise means someone has taken control of keys that can upgrade contracts, change parameters or move funds. This can be very serious, because the attacker may be able to do anything those keys allow. It can also be slower, if the keys are behind a timelock that delays changes.

A frontend or domain hijack means the website has been compromised, often through a DNS takeover or a malicious script injected into the site. The contracts are fine. The danger is that anyone who uses the website is asked to sign malicious transactions. In this case, using the site to “withdraw safely” is exactly the wrong move.

A dependency failure means something the protocol relies on has broken: a bridge, an external lending market, a stablecoin, a price feed operator. The protocol itself may be intact but holding assets that are suddenly worth much less or can’t be moved.

The team’s statement usually makes the category clear. If it doesn’t, the safest assumption is the one that leads to the fewest risky actions: stay away from the website and don’t sign anything until you know more.

Map your own exposure

While you’re confirming, work out exactly what you have at risk. This is where the notes you made when you deposited pay for themselves.

For each protocol you’re in, you should know which wallets hold positions, which contracts those positions are in, roughly how much is in each and whether any wallets have outstanding approvals to the protocol’s contracts. If you don’t have that written down, build it now, quickly, from your wallets and the explorer.

Separate direct exposure from indirect exposure. Direct exposure is funds deposited in the affected contracts. Indirect exposure includes receipt tokens you hold from the protocol, such as vault shares or liquid staking tokens, that other protocols accept, and positions elsewhere that use those tokens as collateral. If a protocol is exploited, its receipt tokens can lose value quickly, which can trigger liquidations in lending markets where they’re used as collateral.

Check approvals separately from deposits. Some exploits abuse approvals rather than deposits. A router or helper contract with a bug might let an attacker pull tokens from any wallet that approved it, including wallets that hold no position at all. If the team says approvals to a particular contract are at risk, that affects every wallet that ever approved it.

For a farm, this can mean a lot of wallets. That’s another argument for keeping a simple record per protocol: wallet, contract, amount, approval. When something breaks, that record turns a panicked hour of searching into ten minutes of checking.

Withdraw only if the path is sound

The instinct in an exploit is to get funds out. Sometimes that’s right. Sometimes it’s pointless, and sometimes it’s dangerous.

Withdrawing makes sense when the team says a specific product is affected and yours isn’t, or when the contract holding your funds is working normally and you want out before anything spreads. In those cases, withdrawing from contracts you’re confident are sound is reasonable.

Withdrawing is pointless when the contract has already been drained or is paused. Many protocols have a pause function for emergencies. If deposits and withdrawals are paused, your transaction will fail, and you’ll pay gas for nothing. That’s not a sign your funds are gone. It’s often a sign the team is trying to stop further damage.

Withdrawing is dangerous when the only path is through a compromised website, or when the incident involves the frontend. If the site has been hijacked, the “withdraw” button may trigger a malicious approval. In that situation, either wait for the team to confirm the site is safe, or withdraw by interacting with the contract directly through the explorer, which you should only do if you’re confident you understand the function you’re calling and the contract address you’re calling it on.

A few practical points. Gas often spikes during incidents, because many people are trying to withdraw at once and attackers may be pushing transactions too. Don’t set absurd fees out of panic. If withdrawing involves a swap, such as exiting a liquidity position into a single asset, keep slippage tight. Loose slippage in a crowded moment invites value extraction by bots.

Withdrawals normally return funds to the wallet that deposited. There’s usually no need to route funds anywhere else in the moment, and rushing to move everything through a single new address adds risk without adding safety.

Receipt tokens and the knock-on effects

Exploits rarely stay inside one protocol. The damage travels through the tokens the protocol issued, and that’s where farmers with layered positions get caught out.

When you deposit into many protocols, you receive a receipt token in return: vault shares, a liquid staking token, an interest-bearing version of the asset you deposited. That receipt is a claim on the protocol’s holdings. If the protocol loses a large share of its holdings, the receipt is suddenly worth less, and the market often prices that in within minutes, sometimes more harshly than the real loss justifies because nobody yet knows the real loss.

The first knock-on effect is liquidations. If you’ve used that receipt token as collateral in a lending market elsewhere, a sharp price drop can push your position below its liquidation threshold. Other markets may use oracles that react quickly, and liquidation bots act faster than people. If you have a borrow position backed by an affected receipt, check it early. Repaying part of the loan or adding different collateral, if you can do so safely, may protect it.

The second is pools. Liquidity pools that pair the receipt token with its underlying asset can become lopsided quickly as holders rush to swap out. If you provide liquidity in such a pool, you may find your position is now mostly the affected token. Withdrawing from a pool during a panic can be sensible, but remember the slippage and fee points above.

The third is dependent protocols. Some protocols build on top of others, holding receipt tokens in their own vaults. If you’re in one of those, you may be exposed to an exploit in a protocol you never used directly. Your per-protocol notes should include what each vault holds underneath, where you can find out.

None of this means you should unwind everything at the first sign of trouble. It means your exposure map needs to include the second layer, not just the first.

When the problem is a dependency

Sometimes the protocol you’re in hasn’t been exploited at all. Something it relies on has.

A bridge failure is the classic case. If a protocol holds bridged assets, and the bridge that issued them is exploited, those assets may no longer be backed. The protocol’s contracts may be flawless and its team blameless, and it’s still holding tokens that have lost much of their value.

Stablecoin trouble works similarly. If a protocol’s vaults are full of a stablecoin that loses its peg, the vault’s value falls with it. That’s covered in more detail in a separate piece on depegs.

Price feed issues, where an oracle provider has problems or a feed stops updating, can cause lending markets to misprice collateral even though nothing in the markets themselves has changed.

In these cases, the protocol’s team may be as surprised as you are, and their early statements may be about the dependency rather than their own code. The actions available to you are often more limited. Withdrawing from the protocol may return the affected asset to you rather than something safer, so withdrawing doesn’t necessarily reduce your exposure. You might just be holding the same problem in your own wallet instead.

It’s still worth checking approvals and staying away from any recovery links. But the decision about whether to act is different. Sometimes the right move is to wait for the dependency’s own team to explain what’s backing the asset now, rather than rushing into a swap at a panicked price.

Talking to other depositors

In the hours after an exploit, the protocol’s community channels fill with worried users. Talking to them can help, and it can also expose you.

Other depositors are often useful sources of information. Someone may notice first that a particular vault is paused, or that withdrawals from another product are working. Just remember they’re guessing too, and verify anything they say against the team’s statements and the explorer.

Be careful what you share. Posting your wallet address to ask “am I affected?” links your chat account to that address in public, invites DMs from fake support agents and gives scammers a target list. If you want to check exposure, do it yourself on the explorer.

Don’t repost links, even ones that look official. A lot of the damage in the second wave comes from well-meaning users sharing a “compensation checker” they found in a reply.

If you see a scam link or a fake announcement, report it to the channel’s moderators rather than engaging with it. And if you see someone else about to click something dangerous, a short, plain warning in public helps more than a DM, which can look like exactly the thing you’re warning about.

Treat speculation carefully. In the first hours, you’ll see confident claims about the size of the loss, whether users will be compensated and who’s responsible. Most of those claims are made without information. It’s fine to read them. It’s rarely wise to act on them.

Revoke approvals to affected contracts

Revoking approvals is often the most useful thing a user can do during an incident, and it’s frequently overlooked because it doesn’t feel like rescuing anything.

An approval lets a contract move tokens from your wallet up to a set amount, often unlimited. If a contract with approval rights is compromised, or has a bug that lets an attacker direct its actions, it can move your tokens even if you never deposited anything into it.

So if the team or credible researchers say a particular contract is involved, revoke any approvals your wallets have granted it. Use a revoke tool you already know, or the approval checker on the chain’s explorer, reached by typing the address rather than clicking a link. Each revoke is a small transaction.

Do this for every wallet that interacted with the protocol, not only the ones with deposits. Farming wallets often approve a router once and forget about it.

Revoking doesn’t recover anything that’s already been taken. It closes a door that might otherwise stay open for as long as the approval exists.

A caution: during incidents, fake revoke sites circulate too. A site claiming to help you “revoke approvals for the exploited protocol” that asks you to sign anything other than a clear revoke transaction is not helping you. A genuine revoke shows up in your wallet as an approval with an amount of zero, or as a clear revoke call. Anything else, stop.

Every significant exploit is followed by a wave of scams aimed at the victims. They’re predictable, and they often do more damage to individual users than the exploit did.

Fake “compensation” and “refund” sites appear within hours. They’re shared in replies to the protocol’s posts, in Discord channels and in DMs. They promise to check whether you’re affected or to let you claim a share of recovered funds. They ask you to connect a wallet and sign.

Fake tokens arrive in wallets that held positions. They’re named something like a refund, a compensation token or a recovery claim, and they point to a site. The mechanics are the same as any unsolicited token: the token is bait, and the site is the attack.

Fake support agents target people asking questions. If you post “I had funds in the exploited vault, what do I do?” in a public channel, expect DMs from accounts pretending to be the team, offering help through a link.

Fake “recovery services” contact people who’ve posted about losses, offering to trace and return stolen funds for an upfront fee.

And compromised or impersonated official accounts may post “migration” instructions pointing to a malicious site.

The rule is short. In the days after an exploit, don’t sign anything on any site you reached from a link. Real compensation plans, if they happen, are announced on the protocol’s official channels, usually after the team has finished investigating and taken a snapshot of affected balances. They aren’t urgent, and they never require a seed phrase or a signature on some new site within the hour.

What the team is likely doing

Understanding what usually happens on the protocol’s side can make the waiting easier and help you judge what they say.

In the first hours, teams typically try to stop the damage. That can mean pausing contracts, disabling a frontend, contacting exchanges and stablecoin issuers to freeze stolen funds where possible, and working with security firms to understand the attack.

Soon after, many teams publish an initial statement acknowledging the incident, saying which products are affected and telling users what to do or avoid. The first statement is often incomplete. Numbers change as the investigation continues.

Some teams attempt to negotiate with the attacker, often by publicly offering a bounty in return for returning most of the funds. This sometimes works, and sometimes funds come back days later. It isn’t something to count on.

Days or weeks later, a post-mortem usually follows, explaining the root cause and what’s being done. If a compensation or recovery plan exists, it’s usually described here or in a separate announcement. These plans vary a lot: some make users whole, some pay out partially, some pay in a new token, some don’t happen at all.

None of that timeline is guaranteed. But if a team is silent for a long time, or its messaging keeps shifting in ways that don’t add up, that’s information about what you can expect.

Document everything while it’s fresh

Whatever happens next, a clear record helps. It helps with any compensation claim, with your own accounting and with deciding what to do differently.

For each affected wallet, record the address, the protocol and contract, the position type and your best estimate of the balance before the incident. If you have a record from before, great. If not, the explorer can usually show historical balances and deposit transactions.

Save the transaction hashes of your deposits and any withdrawals or revokes you made during the incident.

Take screenshots of the protocol’s official announcements, with dates and times. Statements get edited and deleted. Your screenshot records what was said when.

Note the timeline of your own actions: when you heard, when you confirmed, what you did and when.

Keep this in one place. If a compensation process asks for evidence, you’ll have it. If there’s no process, you’ll at least have an accurate record of what you lost, which matters for your own books. How that record should be treated for tax is a question for a professional in your jurisdiction, not something this piece can answer.

Multi-wallet farms in an incident

Farms have a few particular issues during an exploit, and it’s worth thinking about them before one happens.

The biggest is scale. If fifteen wallets each hold a position in the affected protocol, checking, withdrawing and revoking across all of them takes time. Having the record described above, with every wallet’s position in one list, is the difference between a methodical hour and a chaotic afternoon.

Prioritise by value and risk. Deal first with the largest positions in contracts that are still working, then with approvals to contracts named as affected, then with smaller positions.

Keep your process clean. During an incident, it’s tempting to open many tabs, click links from chats and move quickly. That’s exactly when mistakes happen. Use your saved bookmarks and your normal signing setup. Don’t install a new tool or extension someone recommends in the chaos.

Keep funds where withdrawals send them. Withdrawals return to the depositing wallet, and moving everything onward in a rush, especially all into one address, creates new risk and new links between wallets without making anything safer. If you later decide to consolidate, do it deliberately, with the same care you’d use any other time.

And remember that points or eligibility may be affected. Protocols sometimes pause points programs after an incident, or change how they count deposits. Nothing can be predicted here. Focus on the funds first.

A worked example

Here’s a made-up incident to show the order of operations.

An operator has six wallets with deposits in a lending protocol’s stablecoin market and two wallets with deposits in its newer yield vault. All eight approved the protocol’s router at some point.

Late in the evening, a group chat says the protocol is being drained. The operator opens the protocol’s verified X account from a bookmark. There’s a post from ten minutes earlier: an exploit in the new yield vault, core lending markets not affected, vault contracts paused, users should revoke approvals to the router while the team investigates.

The Discord announcements channel says the same. A reply under the X post links a “check your exposure” site. The operator ignores it.

On the explorer, the yield vault shows large outflows ending twenty minutes earlier. The lending market looks normal.

The operator checks their notes. The two vault wallets can’t withdraw because the vault is paused, and trying would just waste gas. The six lending wallets hold positions in contracts the team says are unaffected. The operator decides to withdraw from those, one wallet at a time through the official app, which the team hasn’t flagged as compromised, keeping fees reasonable.

Then the operator revokes router approvals on all eight wallets, using the explorer’s approval checker.

Finally, they record balances, hashes and screenshots, and wait for the post-mortem. Two fake refund tokens arrive in the vault wallets the next day. They’re hidden and ignored.

What people get wrong in the first hour

A few mistakes come up again and again.

Acting on the first post. Early reports are often wrong about which contracts are affected, and acting on them can mean paying gas for failed transactions or withdrawing from something that was fine.

Using the website during a frontend attack. If the site is the problem, the site isn’t the solution.

Clicking links in replies. The replies under an official incident post are prime real estate for scammers.

Raising fees and slippage to extremes. Panic settings rarely help and often cost.

Moving funds somewhere “safe” through a new tool. A brand-new bridge, aggregator or “safe vault” recommended in the chaos is a new risk, not a refuge.

Forgetting approvals. Many users withdraw and consider themselves done, leaving approvals open to a contract that may still be exploitable.

Not writing anything down. Memory is poor under stress, and a compensation process weeks later will want details.

Preparing before it happens

The best time to prepare for an exploit is before you deposit.

Keep a per-protocol record: wallets, contracts, amounts, approvals, and links to the protocol’s official channels. Update it when positions change.

Bookmark the official site, social accounts and status page, so you never need to search during an incident.

Know how to use your explorer’s approval checker, or a revoke tool you trust, before you need it.

Understand, roughly, what each protocol’s emergency powers are: whether it can pause, who controls upgrades, whether there’s a timelock. The piece on reading audit reports covers where to find this.

Size deposits so that a total loss in any one protocol would hurt but not wreck you. Spreading across protocols reduces the damage from any single exploit, though it increases the number of things to monitor.

And decide in advance what you’ll do. If you’ve thought through “if the contract I’m in is paused, I’ll wait” and “if the site is compromised, I won’t use it”, you won’t have to reason it out under pressure.

Where this runbook stops

I want to be clear about what this can and can’t do. In many serious exploits, funds are gone within minutes, and nothing a depositor does afterwards changes that. This runbook is about protecting what’s left, avoiding a second loss and keeping a clear record. It doesn’t promise recovery.

It also can’t tell you whether a particular protocol will compensate users, how much, or when. That depends on the team, the attacker, the size of the loss and many other factors.

I haven’t collected figures on how often compensation happens or how much users typically recover, and I’d be sceptical of anyone quoting a general rate. What the mechanics show is that the first hours reward calm, verified action and punish haste, and most of the extra damage in these incidents comes from the scams that follow rather than the exploit itself.

For more on protocol risk, deposit records and the habits that keep a farm steady when something breaks, head back to the Airdrop Farming homepage.

Get new guides and videos first — join the Telegram channel.

need infra for this today?