Blockchain Vulnerability Classes

Comprehensive coverage of smart contract vulnerability categories. Each class includes detection patterns, real-world exploits, and remediation.

Overview

Class	Exploits	Total Loss	Notable
Reentrancy	47	$890M	The DAO ($60M)
Access Control	89	$1.2B	Ronin ($625M)
Oracle Manipulation	34	$420M	Mango ($114M)
Flash Loan	62	$380M	Cream ($130M)
Logic Errors	156	$950M	Wormhole ($326M)

Reentrancy

Occurs when external calls allow attackers to re-enter the contract before state updates complete.

Vulnerable

Solidity

1function withdraw(uint256 amount) {
2  require(balances[msg.sender] >= amount);
3  // External call BEFORE state update
4  msg.sender.call{value: amount}("");
5  balances[msg.sender] -= amount;
6}

Secure

Solidity

1function withdraw(uint256 amount) {
2  require(balances[msg.sender] >= amount);
3  // State update BEFORE external call
4  balances[msg.sender] -= amount;
5  msg.sender.call{value: amount}("");
6}

Access Control

Missing or flawed permission checks allow unauthorized users to call privileged functions.

Most Common Category

Access control vulnerabilities account for over $1.2B in losses, making them the most costly vulnerability class.

Oracle Manipulation

Price oracles can be manipulated to drain funds from DeFi protocols that rely on them for valuations.

Solidity

1// Vulnerable: Using spot price
2uint256 price = uniswapPair.getReserves();
3
4// Secure: Using TWAP with staleness check
5(, int256 price,, uint256 updatedAt,) = chainlinkFeed.latestRoundData();
6require(block.timestamp - updatedAt < 1 hours, "Stale price");

Flash Loan Attacks

Flash loans enable attackers to borrow massive capital within a single transaction to manipulate markets and exploit protocols.

Attack Pattern

1. Borrow millions via flash loan (no collateral needed)
2. Manipulate on-chain price oracle
3. Exploit protocol using manipulated price
4. Repay flash loan + profit in same transaction

Business Logic Errors

Flaws in the intended business logic that allow unintended behavior. These are the hardest to detect automatically.

AI-Assisted Detection

Bloodhound's AI engine is specifically trained to identify logic errors by understanding the intended behavior of common DeFi patterns.