One (bool ok, ) = msg.sender.call{value: amount}("") placed before balances[msg.sender] = 0, or one initialize() without the initializer modifier, is enough for someone to drain or take over a contract. This skill finds those and nine other Solidity mistakes before you deploy, and tells you how to fix each one. It is not a substitute for a smart contract audit: use it to catch known mistakes early, before an auditor looks at the code.

Your agent runs a scanner on your .sol files. It skips comments, tests, scripts, mocks and dependencies (node_modules, Foundry lib/, Remix .deps), splits each contract into functions, reads their visibility and modifiers, knows the state variables, and follows parameters into delegatecall. It does not flag ordinary OpenZeppelin patterns: Ownable and AccessControl modifiers, ReentrancyGuard nonReentrant, checks-effects-interactions, SafeERC20, Initializable with initializer, ECDSA.recover and Solidity 0.8 checked math. It never compiles, deploys or runs anything. It returns BLOCK, REVIEW, or NOT FLAGGED, with the file, line, reason, and a safer approach.

What it checks

BLOCK (fix before shipping)

• SC01: tx.origin used to authorize (require(tx.origin == owner), admins[tx.origin])

• SC02: delegatecall to an address from the function's parameters or msg.sender, with no access control

• SC03: selfdestruct that anyone can call

• SC04: initialize functions without initializer or another guard, and UUPS _authorizeUpgrade without access control

• SC05: state written after an ETH transfer or low-level call, without nonReentrant (reentrancy)

REVIEW (check before release)

• SC11: floating pragmas (^0.8.20, >=0.8.0) in files with a deployable contract

• SC12: ignored results of call, delegatecall, staticcall and send

• SC13: winners, indexes and outcomes picked from block.prevrandao, block.timestamp or blockhash, also through a local variable (time arithmetic such as % 1 weeks is not reported)

• SC14: compilers before 0.8.0 without SafeMath (silent overflow)

• SC15: mint, set..., pause, upgrade, sweep and withdraw-all functions without onlyOwner, onlyRole or a msg.sender check

• SC16: ecrecover without a check against address(0)

What's included

• The scanner (Python 3.9+, standard library only)

• A rules guide with the reason and safer approach for each rule

• A risky and a safe example Foundry project, and the real report the scanner produces on the risky one

Limits

It reads code as text, not with the Solidity compiler. Inheritance from other files is judged by modifier names, and internal calls, storage pointers, assembly (apart from delegatecall), cross-function and read-only reentrancy, token callbacks, oracle manipulation, front-running and business logic are not checked. It is not a substitute for a smart contract audit. NOT FLAGGED is not a guarantee.

Contents

One ZIP file with 19 files: SKILL.md, the scanner in scripts/ (4 Python files), references/rules.md (every rule with its reason and safer approach), and examples/ (12 example files plus report.md, the real scanner output on the risky example).

How to use: This is an agent skill: a folder with a SKILL.md file and Python scripts that an AI coding agent such as Claude Code, Cursor, or Codex CLI can run. Unzip it into your agent's skills folder (for Claude Code: ~/.claude/skills/), then ask your agent to check your project. The scripts also run on their own with Python 3.9 or newer, with nothing to install. They read your files only: no network access, and nothing is sent anywhere.