Feed

- EIP 7702 Security Considerations - link

- Red Flags and Green Flags of Yield Bearing Stablecoins 

@EthSecurity1

Chainalysis full user data breach: names, usernames, password hashes, valid tokens exposed, some plaintext passwords leaked
@EthSecurity1

BNIX / OSK (CoreVault) hacked for ~$181K

Rootcause: The vault’s BNB principal redemption logic depended on a manipulable on-chain spot price that could be distorted within a single transaction via flash loan.
Key technical failure
redeemPrincipal() calculated the BNB amount to pay out as:
bnbAmount = redeemableUsd × wbnbPerUsd / 1e18
The wbnbPerUsd price came from OracleLib.getAccountingPrices, which used a 3-source median.
When the TWAP / average-price windows were empty or unfilled, the median degraded to the live PancakeSwap WBNB/USDT spot price.
That spot price sits in a low-liquidity pair that is trivial to move with a large flash-loan dump.
Attack flow (simplified)
Attacker flash-borrowed ~207k WBNB.
Dumped ~195k WBNB into the PancakeSwap WBNB/USDT pool → artificially inflated wbnbPerUsd.
Called redeemPrincipal() while the manipulated price was live.
The vault over-paid BNB from the insurance/reserve pool (~260 BNB / ~$181k).
Attacker swapped the excess BNB back, repaid the flash loan, and kept the profit (~160 BNB / ~$112k).
Core design flaw
The protocol treated a same-block, flash-loan-manipulable spot price as a reliable accounting oracle for redemptions. Proper protections (strict TWAP requirement, secondary oracle fallback that cannot degrade to spot, or circuit breakers on large price deviations) were missing or ineffective.

@EthSecurity1

- Preventing Second Preimage Attacks in Merkle Trees: A Complete Guide 

- Stablecoin Security: How Design Choices Create Vulnerabilities and Economic Risk 

-  Inside the FBI: Crypto, Crime & National Security – Ep. 171

@EthSecurity1

SOLANA MEV BOT KING DRAINED OF $786K King, one of Solana's oldest arbitrage bots with $7.2M in lifetime PnL, was drained of ~$786K, including 4.26K $WSOL, 191
$SOL
and $366K in stablecoins. Drainers: BHRayETPjvgg3BCr9H82V1D6NRwxu5v44JAaVuaod3jU Forwarder: Ddmy5oM8xtTs1EBC6eF2SZwQfBreewQpFgtUDm51Hkqj Victim: 7dGrdJRYtsNR8UYxZ3TnifXGjGc9eRYLq9sELwYpuuUu
@EthSecurity1

- When Hackers Get Hacked: Analyzing the Breach of LockBit - link

- The State of Drainers Vol. 1 - link

- Astaroth Banking Trojan Harnessing GitHub to Steal Crypto Credentials.

@EthSecurity1

- RISC Zero Security Disclosure: Arbitrary code execution in guest.

- TON Transaction Simulator for Revealing Smart Contract Vulnerabilities.

@EthSecurity1

Seems chainalysis been hacked


It told customers in an email seen by ICD that an attacker using "offensive agentic techniques" accessed users' names, emails and role labels.

It says this has nothing to do with the $6.3B Binance heist.

@EthSecurity1

Tether and Circle have frozen about $2.1B, including $313M USDT on Tron. Nothing has gone to Tornado Cash. About $718M in ETH and about $500M in USDT are still free to move, and 17,270 BTC sits untouched.
https://arkm.com/explorer/entity/21a94755-f604-45e5-9501-6542e23005e8
@EthSecurity1

Tether freeze ~ 1.9 billion
Total stolen assets beyond $3.5 B
@EthSecurity1

zkLink Nova's Arbitrator hacked for~$780K

Rootcause: The Arbitrator contract had two related flaws that the attacker combined:
Uncleared transient storage of the authentication hash
During a legitimate claim, an authenticated gateway callback stores finalizeMessageHash (in transient storage). The Arbitrator later checks this hash against the expected source gateway, value, and message.
The contract never cleared finalizeMessageHash after use. Transient storage persists for the entire transaction, so a second claim in the same transaction could reuse the first claim’s already-set (and still-valid) hash without triggering a fresh authenticated callback.
Caller-controlled target and calldata in the claim path
The claim function accepted a user-supplied target address and calldata. This let the attacker point the second claim at the Ethereum Gateway and supply sendMessage() calldata that invoked a privileged syncBatchRoot update with an attacker-chosen root.
Resulting attack flow (same transaction):
First claim: legitimate Linea message → sets finalizeMessageHash and forwards normally.
Second claim: reuses the stale hash for verification while forcing the trusted Arbitrator to call the Ethereum Gateway with malicious root-update calldata.
The zkLink contract accepted the attacker-controlled root because it came from the authorized Gateway.
Attacker then submitted matching withdrawal proofs against the forged root and drained funds.
@EthSecurity1

a Safe multisig lost $7.45M due to a misconfiguration on Avalanache.

RootCause: A Safe multisig misconfiguration on Avalanche involving unrevoked unlimited token approvals to a predictable, empty address.The 2-of-7 Safe had given unlimited spending approvals (for WAVAX, USDC, WETH.e, and BTC.b) to the address 0x3725bd4d175283108156c3f15f86e1c51266155d.
On other chains (Base, Arbitrum, Ethereum), that same address held the team’s legitimate pool contract. On Avalanche, however, the address was empty. It is the deterministic address produced by Safe’s public CreateCall contract on its 13th deployment, and CreateCall had never been used on Avalanche. The approvals (set in a batch transaction on 2026-03-07) were never revoked.
Because the target address was unoccupied, anyone could deploy their own contract there (via the Safe singleton factory + CreateCall) and then use the existing unlimited allowances to drain the Safe’s tokens. An attacker did exactly that; an MEV bot front-ran the drain in the same block.
@EthSecurity1

- Device hardening & factory reset guides - link

- Top 5 Crypto & Web3 Hacks That Started With Phishing - link

- Cross-Function Reentrancy: When Functions Betray Each Other 
@EThSecurity1

X 100 hacked for $4.75M

Rootcause : Anyone could instantly join the Senate by freezing X tokens (and paying a small candidacy fee, reported as ~100 X). Membership was based on the current frozen stake. There was no snapshot of balances/voting power and no timelock (or other delay) on proposals.
A proposal could be created, voted on (with the newly acquired majority), and executed in the same transaction, including calls that invoked the treasury’s own swap functions (e.g., v3Swap).
This allowed the attacker to:
1. Flash-borrow USDT (via Aave), buy X tokens, and distribute/freeze them across controlled contracts to seize 2 of the 3 Senate seats.
2. Immediately propose + vote + execute a malicious proposal that forced the Freezer to swap a large amount of treasury X tokens into a freshly created, attacker-controlled (and empty/manipulated) PancakeSwap V3 X-USDT pool with min-output set near zero.
3. Withdraw the liquidity from that pool, sell the recovered X on legitimate markets, repay the flash loan, and keep the profit (~$4.75–4.85M USDT range reported across related alerts).
@EthSecurity1

- The Crypto Threat Landscape: Threats and Exploits Targeting Crypto Users - link

- Blockchain Forensics: A Practical Guide to Tracing Stolen Funds

- Blockchain Forensics: Attribution Techniques and the Role of OSINT

- Blockchain Forensics: Advanced Blockchain Forensics Techniques and Additional Resources
@EthSecurity1

- The adoption of Solana OFAs has increased over the past six months.

to learn about how they work, some issues we hear about from the top Solana apps, and brief thoughts on if these arrangements making Solana better for apps and users. - link

- The sandwich attack vector built to be missed. - link
@EthSecurity1

AlpsCon.ai hacked $152K

Rootcause:
Access-control / operational failure (the entry point)
The Hardhat/Anvil default account #0 (0xf39fd6e51aad88f6f4ce6ab8827279cfffb92266) — whose private key is publicly known — was added to the production KYC whitelist in a bulk import on 2025-08-24 and left there for ~14 months.
Anyone could therefore use EIP-7702 to delegate arbitrary code to that address and execute transactions as a fully verified, whitelisted wallet.
Reward-accounting bug (the actual value extraction)
Newer reward contracts (V7–V9) recursively imported balances from earlier versions. When a fresh stake appeared in an older hub that still accepted deposits, the newer contracts multiplied the newly visible balance by the already-accumulated profitPerShare while the account’s entryProfitPerShare stayed at zero.
This let a brand-new (flash-loan funded) stake claim the entire historical reward history (~$355K USDC), emptying the pool. The attacker then unstaked, sold the ALPS, repaid the flash loan, and kept the net ~$152K.

@EthSecurity1

Trillion Dollar security - link

Validator sniping: How to harvest IP addresses to redirect MEV - link

The Crypto Threat Landscape: Threats and Exploits Targeting Crypto Users
@EthSecurity1

osec_io used AI agents (led by Claude Opus) and custom tooling to decompile GoonFi V2, a closed-source Solana propAMM written in Rust, achieving byte-identical recompilation of all 219 functions from on-chain bytecode. - link

@EthSecurity1

Former Mt. Gox CEO: Ledger Received Hardware Wallet Tampering Reports Since December 2025 but Issued No Public Warning
@EthSecurity1

Here's how exactly Telegram provides user data to U.S. authorities. A letter recently filed in court shows how Telegram transfers that data, such as IP address and phone number.

More: 404media.co/heres-what-tel…
@EthSecurity1

- Red Flags and Green Flags of Yield Bearing Stablecoins - link

- EVM – Cosmos Convergence Research From Security Base Part 1, Part 2, and Part 3 

- Exploring Vulnerabilities and Audit Tips for TON Smart Contracts

@EthSecurity1

- Kocher's Timing Attack, A Journey from Theory to Practice - link

- Blockchain Security — Offense, Defense, and Practices

- Finding Ways To Break Smart Contracts (Auditing: Part 2) - link

@EthSecurity1

@UNCX_token hacked for ≈$87.7K.

Root cause:
VestingManager.release() subtracts claimed shares from global _totalTokenShares but leaves them in the position's _depositedShares.
After release → topUpVesting(), cancel() calculates sharesToRemove from the full lifetime _depositedShares without subtracting schedule released,

so already-released shares are removed from the global total a second time.

The attacker used a direct donation to magnify this stale-share debit,

corrupted the denominator, then released a second attacker-owned vesting to take tokens backing unrelated users.

Attack tx: etherscan.io/tx/0x352065378…
@EthSecurity1

2026 Hardware Wallet Incidents

@EthSecurity1

- How to Manage Crypto Keys Without Losing Sleep - link

- Building Our Own Post-Quantum FIDO Token

- Use mutation testing to find the bugs your tests don't catch 
@EthSecurity1

- Vulnerabilities in Liquity forks Part 1 and Part 2 by VulSight.

- Yo Protocol's Unseen Dangers: Why Code Audits Aren't Enough - link
@EthSecurity1

Papertrade hacked for $ 1.28M

- A quant trader with $13M+ all-time PnL manipulated Papertrade’s oracle by executing massive ETH buys/sells on Hyperliquid to briefly move its price, profiting $1.73M gross on timed long/short flips before a final $450k liquidation left $1.28M net.

- The attack involved 305 trades (296 wins) over 8 hours, with average 98-second holds: open $200M+ Papertrade positions, manipulate Hyperliquid price within seconds, close for profit, and reverse direction repeatedly.
- Papertrade prices directly off Hyperliquid’s ETH spot at the exact second, making it vulnerable to short-term manipulation; the post suggests the team blocked some closes and recommends upgrading to Hyperliquid’s less manipulable official multi-feed oracle.
@EthSecurity1

Papertrade hacked for $ 1.28M
A quant trader with $13M+ all-time PnL manipulated Papertrade’s oracle by executing massive ETH buys/sells on Hyperliquid to briefly move its price, profiting $1.73M gross on timed long/short flips before a final $450k liquidation left $1.28M net.
The attack involved 305 trades (296 wins) over 8 hours, with average 98-second holds: open $200M+ Papertrade positions, manipulate Hyperliquid price within seconds, close for profit, and reverse direction repeatedly.
Papertrade prices directly off Hyperliquid’s ETH spot at the exact second, making it vulnerable to short-term manipulation; the post suggests the team blocked some closes and recommends upgrading to Hyperliquid’s less manipulable official multi-feed oracle.
@EthSecurity1

https://x.com/cayden_liao/status/2109074675695579399
@EthSecurity1