Traditional blockchain systems validate transactions as either valid or invalid but lack any mechanism to monitor suspicious behavioural patterns. This means that nodes can act maliciously whilst remaining technically compliant and undetected. This project addresses that limitation by designing and implementing a decentralised blockchain simulation in Python using a UTXO transaction model and Proof of Work consensus. This is extended with a suspicion scoring system that continuously monitors node behaviour against four defined metrics: transaction frequency, fee behaviour, UTXO fragmentation, and double spend attempts.