@pcaversaccio
> be a lazy dev
> let LLM analyse an Etherscan verified contract
> LLM has code execution capabilities
> LLM listens to the instructions part of the verified source code comments
> get rekt
The future is full of prompt injection attacks. Looking forward to 2026.
ps: this is an illustrative contract I wrote without fancy obfuscation, so pls don't get too much inspired ;)