web3-audit — Claude Code command for smart contract audits
The web3-audit command checks smart contracts against ten bug classes and helps turn a confirmed issue into a Foundry test. Use it for a structured first pass over Solidity code.
MIT ★ 16,193 updated 2026-10-29
/plugin marketplace add elementalsouls/Claude-BugHunter
/plugin install claude-bughunter@elementalsoulsWhat web3-audit does
It starts with screening criteria, then checks ten classes of bugs, including accounting desynchronization, access control, reentrancy, oracle manipulation, signature replay, and proxy issues. A set of grep searches targets files under contracts/; the checklist also gives questions for reviewing the code paths.
For a confirmed issue, the command provides a Foundry PoC template and a test command. The template includes a fork setup, an attacker address, and a balance assertion. Its searches expect a contracts/ directory, so projects with a different layout need adjustments. A template alone does not verify an exploit; the Foundry test has to run.
What it helps with
- check a contract against ten vulnerability classes
- spot early returns that leave accounting state out of sync
- compare access modifiers across related functions
- inspect period, array, and loop boundaries
- draft a Foundry PoC for a confirmed issue
Example prompts for the agent
- Audit
VulnerableContract.solfor the bug classes covered by web3-audit. - Check
contracts/Vault.solfor access-control and reentrancy problems with/web3-audit contracts/Vault.sol. - Create a Foundry PoC template for the confirmed issue in
contracts/Vault.sol.
How to install web3-audit
1Plugin marketplace
/plugin marketplace add elementalsouls/Claude-BugHunter/plugin install claude-bughunter@elementalsouls2Manually: put the files in place
.claude/commands/web3-audit.mdCall it with /web3-audit. For all projects use ~/.claude/commands/
Manually: put the files in place
~/.codex/prompts/web3-audit.mdCall it with /prompts:web3-audit. Codex marks this format deprecated in favour of skills
Manually: put the files in place
.github/prompts/web3-audit.prompt.mdManually: put the files in place
.gemini/commands/web3-audit.tomlGemini stores commands as TOML: the prompt text goes into the prompt field
Questions about web3-audit
How do I install web3-audit in Claude Code?
Run /plugin marketplace add elementalsouls/Claude-BugHunter, then run /plugin install claude-bughunter@elementalsouls.
What does web3-audit do?
It checks Solidity contracts against ten bug classes and provides search commands for reviewing code. For confirmed findings, it supplies a Foundry PoC template.
How can I test whether a finding is reproducible?
Adapt the PoC template to the contract, then run forge test --match-test test_exploit -vvvv. The template itself does not confirm that the exploit works.
Source text
web3-audit.md · MIT
---
name: web3-audit
description: Smart contract security audit — runs through 10 bug class checklist (accounting desync, access control, incomplete path, off-by-one, oracle errors, ERC4626, reentrancy, flash loan, signature replay, proxy/upgrade). Applies pre-dive kill signals first. Generates Foundry PoC template for confirmed findings. Usage: /web3-audit <contract.sol>
---
# /web3-audit
Smart contract security audit using the 10-bug-class methodology.
## Usage
```
/web3-audit VulnerableContract.sol
/web3-audit https://github.com/protocol/contracts
/web3-audit [paste contract code]
```
## Step 0: Pre-Dive Kill Signals
ALWAYS check these BEFORE reading any code:
```
1. TVL < $500K → max payout too low for effort → SKIP
2. 2+ top-tier audits (Halborn, ToB, Cyfrin, OZ) on simple protocol → SKIP
3. Protocol < 500 lines, single A→B→C flow → minimal attack surface → SKIP
4. max_payout = min(10% × TVL, program_cap) → if < $10K → SKIP
Formula: Is [TVL * 10%] > [hours I'll spend * hourly rate]? If not, skip.
```
Only proceed if score >= 6/10:
- TVL > $10M: +2
- Immunefi Critical >= $50K: +2
- No top-tier audit on current version: +2
- < 30 days since deploy: +1
- Protocol you've hunted before: +1
- Upgradeable proxies present: +1
## Step 1: Accounting State Desynchronization (28% of Criticals)
```bash
# Find accounting variables
grep -rn "totalSupply\|totalShares\|totalAssets\|totalDebt\|cumulativeReward" contracts/
# Find ALL early returns in critical functions
grep -rn "\breturn\b" contracts/ -B3 | grep -B3 "if\b"
```
Check: For each early return in claim/redeem/withdraw functions:
- Which state variables are updated in the normal path?
- Are ALL of them also updated in the early return path?
- If A updated but B isn't → potential desync bug
## Step 2: Access Control (19% of Criticals)
```bash
# Sibling function families — do ALL have same modifier set?
grep -rn "function vote\|function poke\|function reset\|function update\|function claim\|function harvest" contracts/ -A2
# Ownership check: existence vs ownership
grep -rn "_requireOwned\|ownerOf\|_isApprovedOrOwner" contracts/ -B5
# Silent modifiers (if without revert)
grep -rn "modifier\b" contracts/ -A8 | grep -B3 "if (" | grep -v "require\|revert"
# Uninitialized proxy
grep -rn "function initialize\b" contracts/ -A3
grep -rn "_disableInitializers()" contracts/
```
Check: Does EVERY sibling function in a family have the SAME modifiers?
## Step 3: Incomplete Code Path (17% of Criticals)
The function family comparison test:
```
1. List all state changes in function A (deposit/place/create)
2. List all state changes in function B (withdraw/update/cancel)
3. For each state change in A: does B have the corresponding reverse?
4. For each token transfer in A: does B have the corresponding refund?
```
```bash
grep -rn "safeApprove\b" contracts/ # safeApprove without zero-reset?
grep -rn "delete\b" contracts/ -B5 # delete before operation completes?
grep -rn "function deposit\|function mint\|function withdraw\|function redeem" contracts/ -A10
```
## Step 4: Off-By-One (22% of Highs)
Mental test: For EVERY `if (A > B)`: "What happens when A == B?" Is that correct?
```bash
# Boundary comparisons
grep -rn "Period\|Epoch\|Deadline\|period\|epoch\|deadline" contracts/ -A3 | grep "[<>][^=]"
# Loop breaks
grep -rn "\bbreak\b" contracts/ -B10
# Array bounds
grep -rn "\.length\s*-\s*1\|i\s*<=\s*.*\.length\b" contracts/
```
## Step 5: Oracle / Price Manipulation
```bash
# Missing staleness check
grep -rn "latestRoundData" contracts/ -A5 | grep -v "updatedAt\|timestamp"
# Pyth confidence interval
grep -rn "getPriceUnsafe\|getPrice\b" contracts/ -A8 | grep -v "conf\|confidence"
# TWAP windows
grep -rn "secondsAgo\|TWAP\|cardinality" contracts/ -A5
```
Check:
- Is staleness checked? (`require(block.timestamp - updatedAt <= MAX_AGE)`)
- Is Pyth confidence interval checked? (`require(conf * 10 <= price)`)
- Is TWAP window > 1800 seconds (30 min)?
## Step 6: ERC4626 Vaults
```bash
grep -rn "function deposit\|function mint\|function withdraw\|function redeem" contracts/ -A10
grep -rn "function transfer\|function transferFrom" contracts/ -A15
```
Check:
- Does `mint()` call the same validation as `deposit()`?
- Does `transfer()` move lock records along with shares?
- Is there a `_decimalsOffset()` virtual shares defense against first-depositor attack?
## Step 7: Reentrancy
```bash
# Effects after interactions
grep -rn "\.call{value\|safeTransfer\|transfer(" contracts/ -B10 | grep -v "require\|revert"
# Missing nonReentrant
grep -rn "function withdraw\|function redeem\|function claim" contracts/ -A2 | grep -v "nonReentrant"
```
Check: Does every function that transfers ETH or ERC20 follow CEI order?
(Checks → Effects → Interactions)
## Step 8: Flash Loan Oracle Manipulation
```bash
grep -rn "getReserves\|getAmountsOut\|slot0\b" contracts/ -A5
```
Check: Any spot price reading from Uniswap reserves/slot0? → flash loan manipulatable.
## Step 9: Signature Replay
```bash
grep -rn "ecrecover\|ECDSA\.recover" contracts/ -B20
grep -rn "nonce\|_nonces\|nonces\[" contracts/
```
Check: Does signed hash include: nonce + chainId + contract address?
## Step 10: Proxy / Upgrade
```bash
grep -rn "function initialize\b\|_disableInitializers\|initializer" contracts/
grep -rn "delegatecall\b" contracts/ -B3
grep -rn "0x360894\|_IMPLEMENTATION_SLOT\|EIP1967" contracts/
```
Check:
- `_disableInitializers()` in implementation constructor?
- Implementation can't be initialized directly by attacker?
- Storage layout compatible between versions?
## Confirming a Finding
Apply the 7-Question Gate:
```
1. Can I demonstrate this with a Foundry test?
2. What is the financial impact (quantify in $)?
3. Is this in the Immunefi scope?
4. Is it a known issue or acknowledged behavior?
5. Does my Foundry PoC actually run? (forge test -vvvv)
```
## Foundry PoC Template
```solidity
// SPDX-License-Identifier: MIT
pragma solidity ^0.8.0;
import "forge-std/Test.sol";
import "../src/VulnerableContract.sol";
contract ExploitTest is Test {
VulnerableContract target;
address attacker = makeAddr("attacker");
function setUp() public {
vm.createSelectFork("mainnet", BLOCK_NUMBER);
target = VulnerableContract(TARGET_ADDRESS);
deal(address(token), attacker, INITIAL_BALANCE);
}
function test_exploit() public {
uint256 before = token.balanceOf(attacker);
vm.startPrank(attacker);
// Execute exploit
vm.stopPrank();
assertGt(token.balanceOf(attacker), before, "Exploit failed");
}
}
```
Run: `forge test --match-test test_exploit -vvvv`
© Sachin Sharma · MIT · elementalsouls/Claude-BugHunter