Four steps: (1) Contract input — paste an Ethereum address (verified source fetched from Etherscan) or raw Solidity. (2) Feature extraction — 53+ security-relevant features are extracted from source code, bytecode, and on-chain metadata. (3) ML classification — an XGBoost model trained on 2,400+ labeled contracts classifies the risk. (4) Explainable results — SHAP feature attribution shows which factors contributed to the score, with no black boxes.
Rule-based detection runs alongside the model to flag critical patterns such as unbounded minting, owner-only sell blocks, and external call risks before the ML score is finalized.
Related pages
SentinelX provides automated security analysis for informational purposes only. Results are not financial advice or guarantees of safety.