This research introduces a novel language model-based approach to detect potentially dangerous coding practices in smart contracts before they lead to security vulnerabilities.

• Identifies and categorizes over 35 specific bad practices in smart contract development
• Proposes SCALM, a specialized LLM framework for smart contract analysis
• Demonstrates how language models can effectively identify code quality issues that increase security risks
• Provides a systematic approach to improving smart contract reliability on the Ethereum platform

For security teams, this research offers a proactive method to identify code patterns that, while not direct vulnerabilities, significantly increase security risk exposure in blockchain applications.

SCALM: Detecting Bad Practices in Smart Contracts Through LLMs