Smart Contract Vulnerabilities logo

Smart Contract Vulnerabilities

OrganizationPopular
yaklang
smart-contract-vulnerabilities

Smart contract vulnerability playbook. Use when auditing Solidity/EVM contracts for reentrancy, integer overflow, access control, delegatecall, flash loan, signature replay, and MEV-related attack patterns.

Overview

Publisheryaklang
Repositoryhack-skills
Skill namesmart-contract-vulnerabilities
Stars
2.3K
Forks
314
Bundled files
1
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by yaklang on GitHub. Read the source before you install it.

Installation

Install the Smart Contract Vulnerabilities AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/yaklang/hack-skills/tree/main/skills/smart-contract-vulnerabilities
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/yaklang/hack-skills.git /tmp/hack-skills
mkdir -p .claude/skills
cp -r /tmp/hack-skills/skills/smart-contract-vulnerabilities .claude/skills/smart-contract-vulnerabilities
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Smart Contract Vulnerabilities in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Smart Contract Vulnerabilities on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Smart Contract Vulnerabilities is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

SKILL: Smart Contract Vulnerabilities — Expert Attack Playbook

AI LOAD INSTRUCTION: Expert smart contract audit techniques. Covers reentrancy (single, cross-function, cross-contract, read-only), integer overflow, access control, delegatecall, randomness manipulation, flash loans, signature replay, front-running/MEV, and CREATE2 exploitation. Base models miss subtle cross-contract reentrancy and storage layout collisions in proxy patterns.

0. RELATED ROUTING

  • defi-attack-patterns when the vulnerability is part of a DeFi protocol exploit (flash loans, oracle manipulation, governance attacks)
  • deserialization-insecure when the target is off-chain infrastructure deserializing blockchain data

Advanced Reference

Also load SOLIDITY_VULN_PATTERNS.md when you need:

  • Side-by-side vulnerable vs fixed code patterns for each vulnerability class
  • Gas optimization traps that introduce vulnerabilities
  • Proxy pattern storage collision examples with slot calculations

1. REENTRANCY

The most iconic smart contract vulnerability. External calls transfer execution control; if state is not updated before the call, the callee can re-enter.

1.1 Classic Reentrancy (Single-Function)

Victim.withdraw()
  ├── checks balance[msg.sender] > 0          ✓
  ├── msg.sender.call{value: balance}("")     ← external call
  │   └── Attacker.receive()
  │       └── Victim.withdraw()               ← re-enters before state update
  │           ├── checks balance[msg.sender]   ← still > 0!
  │           └── sends ETH again
  └── balance[msg.sender] = 0                 ← too late

1.2 Cross-Function Reentrancy

Two functions share state; attacker re-enters a different function during callback:

StepExecutionState
1Call withdraw() → external callbalance still positive
2Attacker fallback calls transfer(attacker2)balance used before reset
3transfer reads stale balance → moves fundsattacker2 receives tokens
4Original withdraw completes, zeroes balancedamage done

1.3 Cross-Contract Reentrancy

Contract A calls Contract B, which calls back into Contract A (or Contract C that reads A's stale state). Especially dangerous in DeFi protocols where multiple contracts share state.

1.4 Read-Only Reentrancy

The re-entered function is a view function used by a third-party contract for price calculation. No state modification in the victim, but the stale intermediate state misleads the reader.

Real-world: Curve pool get_virtual_price() read during remove_liquidity() callback → inflated price → profit on dependent lending protocol.

Mitigations

PatternProtection Level
Checks-Effects-Interactions (CEI)Core defense; update state before external call
ReentrancyGuard (OpenZeppelin)Mutex lock; prevents same-tx re-entry
Pull payment patternEliminate external calls in state-changing functions
CEI + guard on all public functionsDefense-in-depth against cross-function

2. INTEGER OVERFLOW / UNDERFLOW

Pre-Solidity 0.8

Arithmetic silently wraps: uint8(255) + 1 == 0, uint8(0) - 1 == 255.

AttackExample
Balance underflowbalances[attacker] -= amount when amount > balance → huge balance
Supply overflowtotalSupply + mintAmount wraps → bypass cap checks
Timelock bypasslockTime[msg.sender] + extend wraps to past → early unlock

Post-Solidity 0.8

Default checked arithmetic reverts on overflow. But unchecked{} blocks reintroduce risk:

solidity
unchecked {
    // "gas optimization" — but if i can be influenced by user input, overflow returns
    for (uint i = start; i < end; i++) { ... }
}

SafeMath Bypass Scenarios

  • Casting: uint256 → uint128 truncation before SafeMath check
  • Assembly blocks: mstore / add bypass Solidity-level checks
  • Intermediate multiplication overflow before division: (a * b) / c where a * b overflows

3. ACCESS CONTROL

tx.origin vs msg.sender

Propertymsg.sendertx.origin
ValueImmediate callerEOA that initiated the tx
Safe for authYesNo — phishing contract can inherit tx.origin

Attack: trick owner into calling attacker contract → attacker contract calls victim with owner's tx.origin.

Common Patterns

IssueImpact
Missing onlyOwner on critical functionsAnyone can call admin functions
Unprotected selfdestructAnyone can destroy the contract, force-send ETH
Unprotected delegatecallAttacker executes arbitrary code in victim's context
Default visibility (pre-0.6.0)Functions default to public
Missing zero-address checksOwnership transferred to address(0)

4. RANDOMNESS MANIPULATION

On-chain randomness sources are predictable to miners/validators:

SourcePredictability
block.timestampMiner has ~15s window to manipulate
blockhash(block.number - 1)Known to all at execution time
blockhash(block.number)Always returns 0 (current block hash unknown)
block.difficulty / block.prevrandaoPost-merge: known beacon chain value

Commit-reveal bypass: If reveal phase doesn't enforce timeout or bond, attacker can choose not to reveal unfavorable outcomes (selective abort attack).


5. DELEGATECALL VULNERABILITIES

delegatecall executes callee's code in caller's storage context. Storage slot layout must match exactly.

Storage Layout Collision

Proxy (storage):         Implementation (code):
slot 0: owner            slot 0: someVariable
slot 1: implementation   slot 1: anotherVariable

Implementation writes to someVariable (slot 0) → overwrites proxy's owner. Attacker calls implementation function that writes slot 0 → becomes proxy owner.

Function Selector Collision

4-byte function selectors can collide. If proxy's admin() selector collides with implementation's transfer(), calling admin() on the proxy executes transfer() logic.

Tool: cast selectors <bytecode> (Foundry) to enumerate selectors.


6. FRONT-RUNNING / MEV

Transaction Ordering Manipulation

Victim submits DEX swap tx (visible in mempool)
├── Front-runner: buy token before victim (raise price)
├── Victim tx executes at worse price
└── Back-runner: sell token after victim (profit from spread)
= Sandwich attack

Protection Patterns

DefenseMechanism
Commit-revealHide transaction intent until reveal
Flashbots / private mempoolSubmit tx directly to block builder
Slippage protectionSet minAmountOut to limit MEV extraction
Time-lockDelay execution to reduce predictability

7. SIGNATURE REPLAY

Missing Nonce

Reuse a valid signature to repeat the action (e.g., transfer) multiple times.

Cross-Chain Replay

Same contract deployed on multiple chains with same address → signature valid on all chains. Must include block.chainid in signed message.

EIP-712 Implementation Errors

ErrorConsequence
Missing DOMAIN_SEPARATOR with chainIdCross-chain replay
Domain separator cached at deployBreaks after hard fork changing chainId
Missing nonce in struct hashSignature replay
ecrecover returns address(0) on invalid sigPasses == address(0) owner check

8. SELF-DESTRUCT & FORCE-SEND ETH

selfdestruct(recipient) force-sends all contract ETH to recipient — bypasses receive() and fallback(), cannot be rejected.

Breaks contracts that rely on address(this).balance for logic (e.g., require(balance == expected)).

Post-EIP-6780 (Dencun): selfdestruct only sends ETH; code/storage deletion only if called in same tx as creation.


9. CREATE2 & DETERMINISTIC ADDRESS EXPLOITATION

CREATE2 address = keccak256(0xff ++ deployer ++ salt ++ keccak256(initCode)).

AttackMethod
Pre-fund exploitationPredict address → send tokens/ETH before deployment → selfdestruct → redeploy different code at same address
Pre-approve exploitationPredicted address gets token approvals → deploy malicious contract → drain approved tokens
Metamorphic contractsCREATE2 → selfdestruct → CREATE2 with same salt but different initCode (pre-EIP-6780)

10. FLASH LOAN ATTACK PATTERNS

Single transaction:
├── Borrow large amount (no collateral)
├── Manipulate state (price oracle, governance, etc.)
├── Extract profit from manipulated state
├── Repay loan + fee
└── Keep profit

Key: entire sequence must succeed atomically or the whole tx reverts.


11. SHORT ADDRESS ATTACK

EVM pads missing bytes in ABI-encoded calldata with zeros. If transfer(address, uint256) is called with a 19-byte address, the uint256 amount shifts left by 8 bits → multiplied by 256.

Mitigation: validate calldata length; modern Solidity compilers add checks.


12. TOOLS

ToolPurposeUsage
SlitherStatic analysis, vulnerability detectionslither . in project root
MythrilSymbolic execution, path explorationmyth analyze contract.sol
EchidnaProperty-based fuzzingDefine invariants, fuzz for violations
Foundry (Forge)Test framework, fuzzing, gas analysisforge test --fuzz-runs 10000
HardhatDevelopment, testing, deploymentnpx hardhat test
CertoraFormal verificationWrite specs, prove/disprove properties
4naly3erAutomated gas optimization + vuln reportCI integration

13. DECISION TREE

Auditing a smart contract?
├── Is it a proxy pattern?
│   ├── Yes → Check storage layout collision (Section 5)
│   │   ├── Compare slot assignments between proxy and implementation
│   │   ├── Check for function selector collision
│   │   └── Verify initializer cannot be called twice
│   └── No → Continue
├── Does it make external calls?
│   ├── Yes → Check reentrancy (Section 1)
│   │   ├── State updated before call? → CEI pattern OK
│   │   ├── ReentrancyGuard present? → Check all entry points
│   │   ├── Cross-function state sharing? → Cross-function reentrancy risk
│   │   └── View functions read during callback? → Read-only reentrancy
│   └── No → Continue
├── Does it handle tokens/ETH?
│   ├── Yes → Check integer overflow (Section 2)
│   │   ├── Solidity < 0.8? → All arithmetic suspect
│   │   ├── unchecked{} blocks? → Verify no user-influenced values
│   │   └── Casting between uint sizes? → Truncation risk
│   └── Also check self-destruct force-send (Section 8)
├── Does it use signatures?
│   ├── Yes → Check replay (Section 7)
│   │   ├── Nonce included? → Verify incremented
│   │   ├── ChainId included? → Cross-chain safe
│   │   └── ecrecover result checked for address(0)? → OK
│   └── No → Continue
├── Does it use on-chain randomness?
│   ├── Yes → Predictable (Section 4)
│   │   └── Recommend Chainlink VRF or commit-reveal with bond
│   └── No → Continue
├── Does it interact with DeFi protocols?
│   ├── Yes → Load [defi-attack-patterns](../defi-attack-patterns/SKILL.md)
│   │   ├── Flash loan vectors
│   │   ├── Oracle manipulation
│   │   └── MEV exposure
│   └── No → Continue
├── Does it use CREATE2?
│   ├── Yes → Check deterministic address exploitation (Section 9)
│   └── No → Continue
└── Run automated tools (Section 12)
    ├── Slither for static analysis
    ├── Mythril for symbolic execution
    └── Echidna for fuzzing invariants

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Smart Contract Vulnerabilities AI skill do?

Smart contract vulnerability playbook. Use when auditing Solidity/EVM contracts for reentrancy, integer overflow, access control, delegatecall, flash loan, signature replay, and MEV-related attack patterns.

Why use Smart Contract Vulnerabilities on TypingMind?

Because you install it once and use it with any model. Smart Contract Vulnerabilities is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Smart Contract Vulnerabilities in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/yaklang/hack-skills/tree/main/skills/smart-contract-vulnerabilities. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Smart Contract Vulnerabilities?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Smart Contract Vulnerabilities?

As many as you like. As long as a model supports skills, you can use Smart Contract Vulnerabilities with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Smart Contract Vulnerabilities AI skill free?

Yes. It is published on GitHub by yaklang under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇